1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P10003.2/D11.2, except for
4 internationalization features.)
6 Copyright (C) 1993, 1994 Free Software Foundation, Inc.
8 This program is free software; you can redistribute it and/or modify
9 it under the terms of the GNU General Public License as published by
10 the Free Software Foundation; either version 2, or (at your option)
13 This program is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with this program; if not, write to the Free Software
20 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
22 /* AIX requires this to be the first thing in the file. */
23 #if defined (_AIX) && !defined (REGEX_MALLOC)
33 /* We need this for `regex.h', and perhaps for the Emacs include files. */
34 #include <sys/types.h>
36 /* This is for other GNU distributions with internationalized messages.
37 The GNU C Library itself does not yet support such messages. */
41 # define gettext(msgid) (msgid)
44 /* The `emacs' switch turns on certain matching commands
45 that make sense only in Emacs. */
54 /* If we are not linking with Emacs proper,
55 we can't use the relocating allocator
56 even if config.h says that we can. */
66 /* We used to test for `BSTRING' here, but only GCC and Emacs define
67 `BSTRING', as far as I know, and neither of them use this code. */
68 #ifndef INHIBIT_STRING_HEADER
69 #if HAVE_STRING_H || STDC_HEADERS
72 #define bcmp(s1, s2, n) memcmp ((s1), (s2), (n))
75 #define bcopy(s, d, n) memcpy ((d), (s), (n))
78 #define bzero(s, n) memset ((s), 0, (n))
85 /* Define the syntax stuff for \<, \>, etc. */
87 /* This must be nonzero for the wordchar and notwordchar pattern
88 commands in re_match_2. */
93 #ifdef SWITCH_ENUM_BUG
94 #define SWITCH_ENUM_CAST(x) ((int)(x))
96 #define SWITCH_ENUM_CAST(x) (x)
101 extern char *re_syntax_table;
103 #else /* not SYNTAX_TABLE */
105 /* How many characters in the character set. */
106 #define CHAR_SET_SIZE 256
108 static char re_syntax_table[CHAR_SET_SIZE];
119 bzero (re_syntax_table, sizeof re_syntax_table);
121 for (c = 'a'; c <= 'z'; c++)
122 re_syntax_table[c] = Sword;
124 for (c = 'A'; c <= 'Z'; c++)
125 re_syntax_table[c] = Sword;
127 for (c = '0'; c <= '9'; c++)
128 re_syntax_table[c] = Sword;
130 re_syntax_table['_'] = Sword;
135 #endif /* not SYNTAX_TABLE */
137 #define SYNTAX(c) re_syntax_table[c]
139 #endif /* not emacs */
141 /* Get the interface, including the syntax bits. */
144 /* isalpha etc. are used for the character classes. */
147 /* Jim Meyering writes:
149 "... Some ctype macros are valid only for character codes that
150 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
151 using /bin/cc or gcc but without giving an ansi option). So, all
152 ctype uses should be through macros like ISPRINT... If
153 STDC_HEADERS is defined, then autoconf has verified that the ctype
154 macros don't need to be guarded with references to isascii. ...
155 Defining isascii to 1 should let any compiler worth its salt
156 eliminate the && through constant folding." */
158 #if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII))
161 #define ISASCII(c) isascii(c)
165 #define ISBLANK(c) (ISASCII (c) && isblank (c))
167 #define ISBLANK(c) ((c) == ' ' || (c) == '\t')
170 #define ISGRAPH(c) (ISASCII (c) && isgraph (c))
172 #define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
175 #define ISPRINT(c) (ISASCII (c) && isprint (c))
176 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
177 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
178 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
179 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
180 #define ISLOWER(c) (ISASCII (c) && islower (c))
181 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
182 #define ISSPACE(c) (ISASCII (c) && isspace (c))
183 #define ISUPPER(c) (ISASCII (c) && isupper (c))
184 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
190 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
191 since ours (we hope) works properly with all combinations of
192 machines, compilers, `char' and `unsigned char' argument types.
193 (Per Bothner suggested the basic approach.) */
194 #undef SIGN_EXTEND_CHAR
196 #define SIGN_EXTEND_CHAR(c) ((signed char) (c))
197 #else /* not __STDC__ */
198 /* As in Harbison and Steele. */
199 #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
202 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
203 use `alloca' instead of `malloc'. This is because using malloc in
204 re_search* or re_match* could cause memory leaks when C-g is used in
205 Emacs; also, malloc is slower and causes storage fragmentation. On
206 the other hand, malloc is more portable, and easier to debug.
208 Because we sometimes use alloca, some routines have to be macros,
209 not functions -- `alloca'-allocated space disappears at the end of the
210 function it is called in. */
214 #define REGEX_ALLOCATE malloc
215 #define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
216 #define REGEX_FREE free
218 #else /* not REGEX_MALLOC */
220 /* Emacs already defines alloca, sometimes. */
223 /* Make alloca work the best possible way. */
225 #define alloca __builtin_alloca
226 #else /* not __GNUC__ */
229 #else /* not __GNUC__ or HAVE_ALLOCA_H */
230 #ifndef _AIX /* Already did AIX, up at the top. */
232 #endif /* not _AIX */
233 #endif /* not HAVE_ALLOCA_H */
234 #endif /* not __GNUC__ */
236 #endif /* not alloca */
238 #define REGEX_ALLOCATE alloca
240 /* Assumes a `char *destination' variable. */
241 #define REGEX_REALLOCATE(source, osize, nsize) \
242 (destination = (char *) alloca (nsize), \
243 bcopy (source, destination, osize), \
246 /* No need to do anything to free, after alloca. */
247 #define REGEX_FREE(arg) (0)
249 #endif /* not REGEX_MALLOC */
251 /* Define how to allocate the failure stack. */
254 #define REGEX_ALLOCATE_STACK(size) \
255 r_alloc (&failure_stack_ptr, (size))
256 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
257 r_re_alloc (&failure_stack_ptr, (nsize))
258 #define REGEX_FREE_STACK(ptr) \
259 r_alloc_free (&failure_stack_ptr)
261 #else /* not REL_ALLOC */
265 #define REGEX_ALLOCATE_STACK malloc
266 #define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
267 #define REGEX_FREE_STACK free
269 #else /* not REGEX_MALLOC */
271 #define REGEX_ALLOCATE_STACK alloca
273 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
274 REGEX_REALLOCATE (source, osize, nsize)
275 /* No need to explicitly free anything. */
276 #define REGEX_FREE_STACK(arg)
278 #endif /* not REGEX_MALLOC */
279 #endif /* not REL_ALLOC */
282 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
283 `string1' or just past its end. This works if PTR is NULL, which is
285 #define FIRST_STRING_P(ptr) \
286 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
288 /* (Re)Allocate N items of type T using malloc, or fail. */
289 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
290 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
291 #define RETALLOC_IF(addr, n, t) \
292 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
293 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
295 #define BYTEWIDTH 8 /* In bits. */
297 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
301 #define MAX(a, b) ((a) > (b) ? (a) : (b))
302 #define MIN(a, b) ((a) < (b) ? (a) : (b))
304 typedef char boolean;
308 static int re_match_2_internal ();
310 /* These are the command codes that appear in compiled regular
311 expressions. Some opcodes are followed by argument bytes. A
312 command code can specify any interpretation whatsoever for its
313 arguments. Zero bytes may appear in the compiled regular expression. */
319 /* Succeed right away--no more backtracking. */
322 /* Followed by one byte giving n, then by n literal bytes. */
325 /* Matches any (more or less) character. */
328 /* Matches any one char belonging to specified set. First
329 following byte is number of bitmap bytes. Then come bytes
330 for a bitmap saying which chars are in. Bits in each byte
331 are ordered low-bit-first. A character is in the set if its
332 bit is 1. A character too large to have a bit in the map is
333 automatically not in the set. */
336 /* Same parameters as charset, but match any character that is
337 not one of those specified. */
340 /* Start remembering the text that is matched, for storing in a
341 register. Followed by one byte with the register number, in
342 the range 0 to one less than the pattern buffer's re_nsub
343 field. Then followed by one byte with the number of groups
344 inner to this one. (This last has to be part of the
345 start_memory only because we need it in the on_failure_jump
349 /* Stop remembering the text that is matched and store it in a
350 memory register. Followed by one byte with the register
351 number, in the range 0 to one less than `re_nsub' in the
352 pattern buffer, and one byte with the number of inner groups,
353 just like `start_memory'. (We need the number of inner
354 groups here because we don't have any easy way of finding the
355 corresponding start_memory when we're at a stop_memory.) */
358 /* Match a duplicate of something remembered. Followed by one
359 byte containing the register number. */
362 /* Fail unless at beginning of line. */
365 /* Fail unless at end of line. */
368 /* Succeeds if at beginning of buffer (if emacs) or at beginning
369 of string to be matched (if not). */
372 /* Analogously, for end of buffer/string. */
375 /* Followed by two byte relative address to which to jump. */
378 /* Same as jump, but marks the end of an alternative. */
381 /* Followed by two-byte relative address of place to resume at
382 in case of failure. */
385 /* Like on_failure_jump, but pushes a placeholder instead of the
386 current string position when executed. */
387 on_failure_keep_string_jump,
389 /* Throw away latest failure point and then jump to following
390 two-byte relative address. */
393 /* Change to pop_failure_jump if know won't have to backtrack to
394 match; otherwise change to jump. This is used to jump
395 back to the beginning of a repeat. If what follows this jump
396 clearly won't match what the repeat does, such that we can be
397 sure that there is no use backtracking out of repetitions
398 already matched, then we change it to a pop_failure_jump.
399 Followed by two-byte address. */
402 /* Jump to following two-byte address, and push a dummy failure
403 point. This failure point will be thrown away if an attempt
404 is made to use it for a failure. A `+' construct makes this
405 before the first repeat. Also used as an intermediary kind
406 of jump when compiling an alternative. */
409 /* Push a dummy failure point and continue. Used at the end of
413 /* Followed by two-byte relative address and two-byte number n.
414 After matching N times, jump to the address upon failure. */
417 /* Followed by two-byte relative address, and two-byte number n.
418 Jump to the address N times, then fail. */
421 /* Set the following two-byte relative address to the
422 subsequent two-byte number. The address *includes* the two
426 wordchar, /* Matches any word-constituent character. */
427 notwordchar, /* Matches any char that is not a word-constituent. */
429 wordbeg, /* Succeeds if at word beginning. */
430 wordend, /* Succeeds if at word end. */
432 wordbound, /* Succeeds if at a word boundary. */
433 notwordbound /* Succeeds if not at a word boundary. */
436 ,before_dot, /* Succeeds if before point. */
437 at_dot, /* Succeeds if at point. */
438 after_dot, /* Succeeds if after point. */
440 /* Matches any character whose syntax is specified. Followed by
441 a byte which contains a syntax code, e.g., Sword. */
444 /* Matches any character whose syntax is not that specified. */
449 /* Common operations on the compiled pattern. */
451 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
453 #define STORE_NUMBER(destination, number) \
455 (destination)[0] = (number) & 0377; \
456 (destination)[1] = (number) >> 8; \
459 /* Same as STORE_NUMBER, except increment DESTINATION to
460 the byte after where the number is stored. Therefore, DESTINATION
461 must be an lvalue. */
463 #define STORE_NUMBER_AND_INCR(destination, number) \
465 STORE_NUMBER (destination, number); \
466 (destination) += 2; \
469 /* Put into DESTINATION a number stored in two contiguous bytes starting
472 #define EXTRACT_NUMBER(destination, source) \
474 (destination) = *(source) & 0377; \
475 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
480 extract_number (dest, source)
482 unsigned char *source;
484 int temp = SIGN_EXTEND_CHAR (*(source + 1));
485 *dest = *source & 0377;
489 #ifndef EXTRACT_MACROS /* To debug the macros. */
490 #undef EXTRACT_NUMBER
491 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
492 #endif /* not EXTRACT_MACROS */
496 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
497 SOURCE must be an lvalue. */
499 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
501 EXTRACT_NUMBER (destination, source); \
507 extract_number_and_incr (destination, source)
509 unsigned char **source;
511 extract_number (destination, *source);
515 #ifndef EXTRACT_MACROS
516 #undef EXTRACT_NUMBER_AND_INCR
517 #define EXTRACT_NUMBER_AND_INCR(dest, src) \
518 extract_number_and_incr (&dest, &src)
519 #endif /* not EXTRACT_MACROS */
523 /* If DEBUG is defined, Regex prints many voluminous messages about what
524 it is doing (if the variable `debug' is nonzero). If linked with the
525 main program in `iregex.c', you can enter patterns and strings
526 interactively. And if linked with the main program in `main.c' and
527 the other test files, you can run the already-written tests. */
531 /* We use standard I/O for debugging. */
534 /* It is useful to test things that ``must'' be true when debugging. */
537 static int debug = 0;
539 #define DEBUG_STATEMENT(e) e
540 #define DEBUG_PRINT1(x) if (debug) printf (x)
541 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
542 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
543 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
544 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
545 if (debug) print_partial_compiled_pattern (s, e)
546 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
547 if (debug) print_double_string (w, s1, sz1, s2, sz2)
550 /* Print the fastmap in human-readable form. */
553 print_fastmap (fastmap)
556 unsigned was_a_range = 0;
559 while (i < (1 << BYTEWIDTH))
565 while (i < (1 << BYTEWIDTH) && fastmap[i])
581 /* Print a compiled pattern string in human-readable form, starting at
582 the START pointer into it and ending just before the pointer END. */
585 print_partial_compiled_pattern (start, end)
586 unsigned char *start;
590 unsigned char *p = start;
591 unsigned char *pend = end;
599 /* Loop over pattern commands. */
602 printf ("%d:\t", p - start);
604 switch ((re_opcode_t) *p++)
612 printf ("/exactn/%d", mcnt);
623 printf ("/start_memory/%d/%d", mcnt, *p++);
628 printf ("/stop_memory/%d/%d", mcnt, *p++);
632 printf ("/duplicate/%d", *p++);
642 register int c, last = -100;
643 register int in_range = 0;
645 printf ("/charset [%s",
646 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
648 assert (p + *p < pend);
650 for (c = 0; c < 256; c++)
652 && (p[1 + (c/8)] & (1 << (c % 8))))
654 /* Are we starting a range? */
655 if (last + 1 == c && ! in_range)
660 /* Have we broken a range? */
661 else if (last + 1 != c && in_range)
690 case on_failure_jump:
691 extract_number_and_incr (&mcnt, &p);
692 printf ("/on_failure_jump to %d", p + mcnt - start);
695 case on_failure_keep_string_jump:
696 extract_number_and_incr (&mcnt, &p);
697 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
700 case dummy_failure_jump:
701 extract_number_and_incr (&mcnt, &p);
702 printf ("/dummy_failure_jump to %d", p + mcnt - start);
705 case push_dummy_failure:
706 printf ("/push_dummy_failure");
710 extract_number_and_incr (&mcnt, &p);
711 printf ("/maybe_pop_jump to %d", p + mcnt - start);
714 case pop_failure_jump:
715 extract_number_and_incr (&mcnt, &p);
716 printf ("/pop_failure_jump to %d", p + mcnt - start);
720 extract_number_and_incr (&mcnt, &p);
721 printf ("/jump_past_alt to %d", p + mcnt - start);
725 extract_number_and_incr (&mcnt, &p);
726 printf ("/jump to %d", p + mcnt - start);
730 extract_number_and_incr (&mcnt, &p);
731 extract_number_and_incr (&mcnt2, &p);
732 printf ("/succeed_n to %d, %d times", p + mcnt - start, mcnt2);
736 extract_number_and_incr (&mcnt, &p);
737 extract_number_and_incr (&mcnt2, &p);
738 printf ("/jump_n to %d, %d times", p + mcnt - start, mcnt2);
742 extract_number_and_incr (&mcnt, &p);
743 extract_number_and_incr (&mcnt2, &p);
744 printf ("/set_number_at location %d to %d", p + mcnt - start, mcnt2);
748 printf ("/wordbound");
752 printf ("/notwordbound");
764 printf ("/before_dot");
772 printf ("/after_dot");
776 printf ("/syntaxspec");
778 printf ("/%d", mcnt);
782 printf ("/notsyntaxspec");
784 printf ("/%d", mcnt);
789 printf ("/wordchar");
793 printf ("/notwordchar");
805 printf ("?%d", *(p-1));
811 printf ("%d:\tend of pattern.\n", p - start);
816 print_compiled_pattern (bufp)
817 struct re_pattern_buffer *bufp;
819 unsigned char *buffer = bufp->buffer;
821 print_partial_compiled_pattern (buffer, buffer + bufp->used);
822 printf ("%d bytes used/%d bytes allocated.\n", bufp->used, bufp->allocated);
824 if (bufp->fastmap_accurate && bufp->fastmap)
826 printf ("fastmap: ");
827 print_fastmap (bufp->fastmap);
830 printf ("re_nsub: %d\t", bufp->re_nsub);
831 printf ("regs_alloc: %d\t", bufp->regs_allocated);
832 printf ("can_be_null: %d\t", bufp->can_be_null);
833 printf ("newline_anchor: %d\n", bufp->newline_anchor);
834 printf ("no_sub: %d\t", bufp->no_sub);
835 printf ("not_bol: %d\t", bufp->not_bol);
836 printf ("not_eol: %d\t", bufp->not_eol);
837 printf ("syntax: %d\n", bufp->syntax);
838 /* Perhaps we should print the translate table? */
843 print_double_string (where, string1, size1, string2, size2)
856 if (FIRST_STRING_P (where))
858 for (this_char = where - string1; this_char < size1; this_char++)
859 putchar (string1[this_char]);
864 for (this_char = where - string2; this_char < size2; this_char++)
865 putchar (string2[this_char]);
869 #else /* not DEBUG */
874 #define DEBUG_STATEMENT(e)
875 #define DEBUG_PRINT1(x)
876 #define DEBUG_PRINT2(x1, x2)
877 #define DEBUG_PRINT3(x1, x2, x3)
878 #define DEBUG_PRINT4(x1, x2, x3, x4)
879 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
880 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
882 #endif /* not DEBUG */
884 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
885 also be assigned to arbitrarily: each pattern buffer stores its own
886 syntax, so it can be changed between regex compilations. */
887 /* This has no initializer because initialized variables in Emacs
888 become read-only after dumping. */
889 reg_syntax_t re_syntax_options;
892 /* Specify the precise syntax of regexps for compilation. This provides
893 for compatibility for various utilities which historically have
894 different, incompatible syntaxes.
896 The argument SYNTAX is a bit mask comprised of the various bits
897 defined in regex.h. We return the old syntax. */
900 re_set_syntax (syntax)
903 reg_syntax_t ret = re_syntax_options;
905 re_syntax_options = syntax;
909 /* This table gives an error message for each of the error codes listed
910 in regex.h. Obviously the order here has to be same as there.
911 POSIX doesn't require that we do anything for REG_NOERROR,
912 but why not be nice? */
914 static const char *re_error_msgid[] =
915 { "Success", /* REG_NOERROR */
916 "No match", /* REG_NOMATCH */
917 "Invalid regular expression", /* REG_BADPAT */
918 "Invalid collation character", /* REG_ECOLLATE */
919 "Invalid character class name", /* REG_ECTYPE */
920 "Trailing backslash", /* REG_EESCAPE */
921 "Invalid back reference", /* REG_ESUBREG */
922 "Unmatched [ or [^", /* REG_EBRACK */
923 "Unmatched ( or \\(", /* REG_EPAREN */
924 "Unmatched \\{", /* REG_EBRACE */
925 "Invalid content of \\{\\}", /* REG_BADBR */
926 "Invalid range end", /* REG_ERANGE */
927 "Memory exhausted", /* REG_ESPACE */
928 "Invalid preceding regular expression", /* REG_BADRPT */
929 "Premature end of regular expression", /* REG_EEND */
930 "Regular expression too big", /* REG_ESIZE */
931 "Unmatched ) or \\)", /* REG_ERPAREN */
934 /* Avoiding alloca during matching, to placate r_alloc. */
936 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
937 searching and matching functions should not call alloca. On some
938 systems, alloca is implemented in terms of malloc, and if we're
939 using the relocating allocator routines, then malloc could cause a
940 relocation, which might (if the strings being searched are in the
941 ralloc heap) shift the data out from underneath the regexp
944 Here's another reason to avoid allocation: Emacs
945 processes input from X in a signal handler; processing X input may
946 call malloc; if input arrives while a matching routine is calling
947 malloc, then we're scrod. But Emacs can't just block input while
948 calling matching routines; then we don't notice interrupts when
949 they come in. So, Emacs blocks input around all regexp calls
950 except the matching calls, which it leaves unprotected, in the
951 faith that they will not malloc. */
953 /* Normally, this is fine. */
954 #define MATCH_MAY_ALLOCATE
956 /* When using GNU C, we are not REALLY using the C alloca, no matter
957 what config.h may say. So don't take precautions for it. */
962 /* The match routines may not allocate if (1) they would do it with malloc
963 and (2) it's not safe for them to use malloc.
964 Note that if REL_ALLOC is defined, matching would not use malloc for the
965 failure stack, but we would still use it for the register vectors;
966 so REL_ALLOC should not affect this. */
967 #if (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (emacs)
968 #undef MATCH_MAY_ALLOCATE
972 /* Failure stack declarations and macros; both re_compile_fastmap and
973 re_match_2 use a failure stack. These have to be macros because of
974 REGEX_ALLOCATE_STACK. */
977 /* Number of failure points for which to initially allocate space
978 when matching. If this number is exceeded, we allocate more
979 space, so it is not a hard limit. */
980 #ifndef INIT_FAILURE_ALLOC
981 #define INIT_FAILURE_ALLOC 5
984 /* Roughly the maximum number of failure points on the stack. Would be
985 exactly that if always used MAX_FAILURE_SPACE each time we failed.
986 This is a variable only so users of regex can assign to it; we never
987 change it ourselves. */
988 #if defined (MATCH_MAY_ALLOCATE)
989 int re_max_failures = 200000;
991 int re_max_failures = 2000;
996 unsigned char *pointer;
1000 typedef union fail_stack_elt fail_stack_elt_t;
1004 fail_stack_elt_t *stack;
1006 unsigned avail; /* Offset of next open position. */
1009 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1010 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1011 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1014 /* Initialize `fail_stack'. Do `return -2' if the alloc fails. */
1016 #ifdef MATCH_MAY_ALLOCATE
1017 #define INIT_FAIL_STACK() \
1019 fail_stack.stack = (fail_stack_elt_t *) \
1020 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1022 if (fail_stack.stack == NULL) \
1025 fail_stack.size = INIT_FAILURE_ALLOC; \
1026 fail_stack.avail = 0; \
1029 #define INIT_FAIL_STACK() \
1031 fail_stack.avail = 0; \
1036 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1038 Return 1 if succeeds, and 0 if either ran out of memory
1039 allocating space for it or it was already too large.
1041 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1043 #define DOUBLE_FAIL_STACK(fail_stack) \
1044 ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \
1046 : ((fail_stack).stack = (fail_stack_elt_t *) \
1047 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1048 (fail_stack).size * sizeof (fail_stack_elt_t), \
1049 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1051 (fail_stack).stack == NULL \
1053 : ((fail_stack).size <<= 1, \
1057 /* Push pointer POINTER on FAIL_STACK.
1058 Return 1 if was able to do so and 0 if ran out of memory allocating
1060 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1061 ((FAIL_STACK_FULL () \
1062 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1064 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1067 /* Push a pointer value onto the failure stack.
1068 Assumes the variable `fail_stack'. Probably should only
1069 be called from within `PUSH_FAILURE_POINT'. */
1070 #define PUSH_FAILURE_POINTER(item) \
1071 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1073 /* This pushes an integer-valued item onto the failure stack.
1074 Assumes the variable `fail_stack'. Probably should only
1075 be called from within `PUSH_FAILURE_POINT'. */
1076 #define PUSH_FAILURE_INT(item) \
1077 fail_stack.stack[fail_stack.avail++].integer = (item)
1079 /* Push a fail_stack_elt_t value onto the failure stack.
1080 Assumes the variable `fail_stack'. Probably should only
1081 be called from within `PUSH_FAILURE_POINT'. */
1082 #define PUSH_FAILURE_ELT(item) \
1083 fail_stack.stack[fail_stack.avail++] = (item)
1085 /* These three POP... operations complement the three PUSH... operations.
1086 All assume that `fail_stack' is nonempty. */
1087 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1088 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1089 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1091 /* Used to omit pushing failure point id's when we're not debugging. */
1093 #define DEBUG_PUSH PUSH_FAILURE_INT
1094 #define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1096 #define DEBUG_PUSH(item)
1097 #define DEBUG_POP(item_addr)
1101 /* Push the information about the state we will need
1102 if we ever fail back to it.
1104 Requires variables fail_stack, regstart, regend, reg_info, and
1105 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be
1108 Does `return FAILURE_CODE' if runs out of memory. */
1110 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1112 char *destination; \
1113 /* Must be int, so when we don't save any registers, the arithmetic \
1114 of 0 + -1 isn't done as unsigned. */ \
1117 DEBUG_STATEMENT (failure_id++); \
1118 DEBUG_STATEMENT (nfailure_points_pushed++); \
1119 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1120 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1121 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1123 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \
1124 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1126 /* Ensure we have enough space allocated for what we will push. */ \
1127 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1129 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1130 return failure_code; \
1132 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1133 (fail_stack).size); \
1134 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1137 /* Push the info, starting with the registers. */ \
1138 DEBUG_PRINT1 ("\n"); \
1140 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1143 DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \
1144 DEBUG_STATEMENT (num_regs_pushed++); \
1146 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1147 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1149 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1150 PUSH_FAILURE_POINTER (regend[this_reg]); \
1152 DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \
1153 DEBUG_PRINT2 (" match_null=%d", \
1154 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1155 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1156 DEBUG_PRINT2 (" matched_something=%d", \
1157 MATCHED_SOMETHING (reg_info[this_reg])); \
1158 DEBUG_PRINT2 (" ever_matched=%d", \
1159 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1160 DEBUG_PRINT1 ("\n"); \
1161 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1164 DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\
1165 PUSH_FAILURE_INT (lowest_active_reg); \
1167 DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\
1168 PUSH_FAILURE_INT (highest_active_reg); \
1170 DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \
1171 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1172 PUSH_FAILURE_POINTER (pattern_place); \
1174 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \
1175 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1177 DEBUG_PRINT1 ("'\n"); \
1178 PUSH_FAILURE_POINTER (string_place); \
1180 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1181 DEBUG_PUSH (failure_id); \
1184 /* This is the number of items that are pushed and popped on the stack
1185 for each register. */
1186 #define NUM_REG_ITEMS 3
1188 /* Individual items aside from the registers. */
1190 #define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1192 #define NUM_NONREG_ITEMS 4
1195 /* We push at most this many items on the stack. */
1196 #define MAX_FAILURE_ITEMS ((num_regs - 1) * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1198 /* We actually push this many items. */
1199 #define NUM_FAILURE_ITEMS \
1200 ((highest_active_reg - lowest_active_reg + 1) * NUM_REG_ITEMS \
1203 /* How many items can still be added to the stack without overflowing it. */
1204 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1207 /* Pops what PUSH_FAIL_STACK pushes.
1209 We restore into the parameters, all of which should be lvalues:
1210 STR -- the saved data position.
1211 PAT -- the saved pattern position.
1212 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1213 REGSTART, REGEND -- arrays of string positions.
1214 REG_INFO -- array of information about each subexpression.
1216 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1217 `pend', `string1', `size1', `string2', and `size2'. */
1219 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1221 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \
1223 const unsigned char *string_temp; \
1225 assert (!FAIL_STACK_EMPTY ()); \
1227 /* Remove failure points and point to how many regs pushed. */ \
1228 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1229 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1230 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1232 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1234 DEBUG_POP (&failure_id); \
1235 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1237 /* If the saved string location is NULL, it came from an \
1238 on_failure_keep_string_jump opcode, and we want to throw away the \
1239 saved NULL, thus retaining our current position in the string. */ \
1240 string_temp = POP_FAILURE_POINTER (); \
1241 if (string_temp != NULL) \
1242 str = (const char *) string_temp; \
1244 DEBUG_PRINT2 (" Popping string 0x%x: `", str); \
1245 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1246 DEBUG_PRINT1 ("'\n"); \
1248 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1249 DEBUG_PRINT2 (" Popping pattern 0x%x: ", pat); \
1250 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1252 /* Restore register info. */ \
1253 high_reg = (unsigned) POP_FAILURE_INT (); \
1254 DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \
1256 low_reg = (unsigned) POP_FAILURE_INT (); \
1257 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \
1259 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1261 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \
1263 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1264 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \
1266 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1267 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1269 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1270 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1273 set_regs_matched_done = 0; \
1274 DEBUG_STATEMENT (nfailure_points_popped++); \
1275 } /* POP_FAILURE_POINT */
1279 /* Structure for per-register (a.k.a. per-group) information.
1280 Other register information, such as the
1281 starting and ending positions (which are addresses), and the list of
1282 inner groups (which is a bits list) are maintained in separate
1285 We are making a (strictly speaking) nonportable assumption here: that
1286 the compiler will pack our bit fields into something that fits into
1287 the type of `word', i.e., is something that fits into one item on the
1292 fail_stack_elt_t word;
1295 /* This field is one if this group can match the empty string,
1296 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1297 #define MATCH_NULL_UNSET_VALUE 3
1298 unsigned match_null_string_p : 2;
1299 unsigned is_active : 1;
1300 unsigned matched_something : 1;
1301 unsigned ever_matched_something : 1;
1303 } register_info_type;
1305 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1306 #define IS_ACTIVE(R) ((R).bits.is_active)
1307 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1308 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1311 /* Call this when have matched a real character; it sets `matched' flags
1312 for the subexpressions which we are currently inside. Also records
1313 that those subexprs have matched. */
1314 #define SET_REGS_MATCHED() \
1317 if (!set_regs_matched_done) \
1320 set_regs_matched_done = 1; \
1321 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1323 MATCHED_SOMETHING (reg_info[r]) \
1324 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1331 /* Registers are set to a sentinel when they haven't yet matched. */
1332 static char reg_unset_dummy;
1333 #define REG_UNSET_VALUE (®_unset_dummy)
1334 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1336 /* Subroutine declarations and macros for regex_compile. */
1338 static void store_op1 (), store_op2 ();
1339 static void insert_op1 (), insert_op2 ();
1340 static boolean at_begline_loc_p (), at_endline_loc_p ();
1341 static boolean group_in_compile_stack ();
1342 static reg_errcode_t compile_range ();
1344 /* Fetch the next character in the uncompiled pattern---translating it
1345 if necessary. Also cast from a signed character in the constant
1346 string passed to us by the user to an unsigned char that we can use
1347 as an array index (in, e.g., `translate'). */
1348 #define PATFETCH(c) \
1349 do {if (p == pend) return REG_EEND; \
1350 c = (unsigned char) *p++; \
1351 if (translate) c = translate[c]; \
1354 /* Fetch the next character in the uncompiled pattern, with no
1356 #define PATFETCH_RAW(c) \
1357 do {if (p == pend) return REG_EEND; \
1358 c = (unsigned char) *p++; \
1361 /* Go backwards one character in the pattern. */
1362 #define PATUNFETCH p--
1365 /* If `translate' is non-null, return translate[D], else just D. We
1366 cast the subscript to translate because some data is declared as
1367 `char *', to avoid warnings when a string constant is passed. But
1368 when we use a character as a subscript we must make it unsigned. */
1369 #define TRANSLATE(d) (translate ? translate[(unsigned char) (d)] : (d))
1372 /* Macros for outputting the compiled pattern into `buffer'. */
1374 /* If the buffer isn't allocated when it comes in, use this. */
1375 #define INIT_BUF_SIZE 32
1377 /* Make sure we have at least N more bytes of space in buffer. */
1378 #define GET_BUFFER_SPACE(n) \
1379 while (b - bufp->buffer + (n) > bufp->allocated) \
1382 /* Make sure we have one more byte of buffer space and then add C to it. */
1383 #define BUF_PUSH(c) \
1385 GET_BUFFER_SPACE (1); \
1386 *b++ = (unsigned char) (c); \
1390 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1391 #define BUF_PUSH_2(c1, c2) \
1393 GET_BUFFER_SPACE (2); \
1394 *b++ = (unsigned char) (c1); \
1395 *b++ = (unsigned char) (c2); \
1399 /* As with BUF_PUSH_2, except for three bytes. */
1400 #define BUF_PUSH_3(c1, c2, c3) \
1402 GET_BUFFER_SPACE (3); \
1403 *b++ = (unsigned char) (c1); \
1404 *b++ = (unsigned char) (c2); \
1405 *b++ = (unsigned char) (c3); \
1409 /* Store a jump with opcode OP at LOC to location TO. We store a
1410 relative address offset by the three bytes the jump itself occupies. */
1411 #define STORE_JUMP(op, loc, to) \
1412 store_op1 (op, loc, (to) - (loc) - 3)
1414 /* Likewise, for a two-argument jump. */
1415 #define STORE_JUMP2(op, loc, to, arg) \
1416 store_op2 (op, loc, (to) - (loc) - 3, arg)
1418 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1419 #define INSERT_JUMP(op, loc, to) \
1420 insert_op1 (op, loc, (to) - (loc) - 3, b)
1422 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1423 #define INSERT_JUMP2(op, loc, to, arg) \
1424 insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
1427 /* This is not an arbitrary limit: the arguments which represent offsets
1428 into the pattern are two bytes long. So if 2^16 bytes turns out to
1429 be too small, many things would have to change. */
1430 #define MAX_BUF_SIZE (1L << 16)
1433 /* Extend the buffer by twice its current size via realloc and
1434 reset the pointers that pointed into the old block to point to the
1435 correct places in the new one. If extending the buffer results in it
1436 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1437 #define EXTEND_BUFFER() \
1439 unsigned char *old_buffer = bufp->buffer; \
1440 if (bufp->allocated == MAX_BUF_SIZE) \
1442 bufp->allocated <<= 1; \
1443 if (bufp->allocated > MAX_BUF_SIZE) \
1444 bufp->allocated = MAX_BUF_SIZE; \
1445 bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\
1446 if (bufp->buffer == NULL) \
1447 return REG_ESPACE; \
1448 /* If the buffer moved, move all the pointers into it. */ \
1449 if (old_buffer != bufp->buffer) \
1451 b = (b - old_buffer) + bufp->buffer; \
1452 begalt = (begalt - old_buffer) + bufp->buffer; \
1453 if (fixup_alt_jump) \
1454 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1456 laststart = (laststart - old_buffer) + bufp->buffer; \
1457 if (pending_exact) \
1458 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1463 /* Since we have one byte reserved for the register number argument to
1464 {start,stop}_memory, the maximum number of groups we can report
1465 things about is what fits in that byte. */
1466 #define MAX_REGNUM 255
1468 /* But patterns can have more than `MAX_REGNUM' registers. We just
1469 ignore the excess. */
1470 typedef unsigned regnum_t;
1473 /* Macros for the compile stack. */
1475 /* Since offsets can go either forwards or backwards, this type needs to
1476 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1477 typedef int pattern_offset_t;
1481 pattern_offset_t begalt_offset;
1482 pattern_offset_t fixup_alt_jump;
1483 pattern_offset_t inner_group_offset;
1484 pattern_offset_t laststart_offset;
1486 } compile_stack_elt_t;
1491 compile_stack_elt_t *stack;
1493 unsigned avail; /* Offset of next open position. */
1494 } compile_stack_type;
1497 #define INIT_COMPILE_STACK_SIZE 32
1499 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1500 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1502 /* The next available element. */
1503 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1506 /* Set the bit for character C in a list. */
1507 #define SET_LIST_BIT(c) \
1508 (b[((unsigned char) (c)) / BYTEWIDTH] \
1509 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1512 /* Get the next unsigned number in the uncompiled pattern. */
1513 #define GET_UNSIGNED_NUMBER(num) \
1517 while (ISDIGIT (c)) \
1521 num = num * 10 + c - '0'; \
1529 #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1531 #define IS_CHAR_CLASS(string) \
1532 (STREQ (string, "alpha") || STREQ (string, "upper") \
1533 || STREQ (string, "lower") || STREQ (string, "digit") \
1534 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1535 || STREQ (string, "space") || STREQ (string, "print") \
1536 || STREQ (string, "punct") || STREQ (string, "graph") \
1537 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1539 #ifndef MATCH_MAY_ALLOCATE
1541 /* If we cannot allocate large objects within re_match_2_internal,
1542 we make the fail stack and register vectors global.
1543 The fail stack, we grow to the maximum size when a regexp
1545 The register vectors, we adjust in size each time we
1546 compile a regexp, according to the number of registers it needs. */
1548 static fail_stack_type fail_stack;
1550 /* Size with which the following vectors are currently allocated.
1551 That is so we can make them bigger as needed,
1552 but never make them smaller. */
1553 static int regs_allocated_size;
1555 static const char ** regstart, ** regend;
1556 static const char ** old_regstart, ** old_regend;
1557 static const char **best_regstart, **best_regend;
1558 static register_info_type *reg_info;
1559 static const char **reg_dummy;
1560 static register_info_type *reg_info_dummy;
1562 /* Make the register vectors big enough for NUM_REGS registers,
1563 but don't make them smaller. */
1566 regex_grow_registers (num_regs)
1569 if (num_regs > regs_allocated_size)
1571 RETALLOC_IF (regstart, num_regs, const char *);
1572 RETALLOC_IF (regend, num_regs, const char *);
1573 RETALLOC_IF (old_regstart, num_regs, const char *);
1574 RETALLOC_IF (old_regend, num_regs, const char *);
1575 RETALLOC_IF (best_regstart, num_regs, const char *);
1576 RETALLOC_IF (best_regend, num_regs, const char *);
1577 RETALLOC_IF (reg_info, num_regs, register_info_type);
1578 RETALLOC_IF (reg_dummy, num_regs, const char *);
1579 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1581 regs_allocated_size = num_regs;
1585 #endif /* not MATCH_MAY_ALLOCATE */
1587 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1588 Returns one of error codes defined in `regex.h', or zero for success.
1590 Assumes the `allocated' (and perhaps `buffer') and `translate'
1591 fields are set in BUFP on entry.
1593 If it succeeds, results are put in BUFP (if it returns an error, the
1594 contents of BUFP are undefined):
1595 `buffer' is the compiled pattern;
1596 `syntax' is set to SYNTAX;
1597 `used' is set to the length of the compiled pattern;
1598 `fastmap_accurate' is zero;
1599 `re_nsub' is the number of subexpressions in PATTERN;
1600 `not_bol' and `not_eol' are zero;
1602 The `fastmap' and `newline_anchor' fields are neither
1603 examined nor set. */
1605 /* Return, freeing storage we allocated. */
1606 #define FREE_STACK_RETURN(value) \
1607 return (free (compile_stack.stack), value)
1609 static reg_errcode_t
1610 regex_compile (pattern, size, syntax, bufp)
1611 const char *pattern;
1613 reg_syntax_t syntax;
1614 struct re_pattern_buffer *bufp;
1616 /* We fetch characters from PATTERN here. Even though PATTERN is
1617 `char *' (i.e., signed), we declare these variables as unsigned, so
1618 they can be reliably used as array indices. */
1619 register unsigned char c, c1;
1621 /* A random temporary spot in PATTERN. */
1624 /* Points to the end of the buffer, where we should append. */
1625 register unsigned char *b;
1627 /* Keeps track of unclosed groups. */
1628 compile_stack_type compile_stack;
1630 /* Points to the current (ending) position in the pattern. */
1631 const char *p = pattern;
1632 const char *pend = pattern + size;
1634 /* How to translate the characters in the pattern. */
1635 char *translate = bufp->translate;
1637 /* Address of the count-byte of the most recently inserted `exactn'
1638 command. This makes it possible to tell if a new exact-match
1639 character can be added to that command or if the character requires
1640 a new `exactn' command. */
1641 unsigned char *pending_exact = 0;
1643 /* Address of start of the most recently finished expression.
1644 This tells, e.g., postfix * where to find the start of its
1645 operand. Reset at the beginning of groups and alternatives. */
1646 unsigned char *laststart = 0;
1648 /* Address of beginning of regexp, or inside of last group. */
1649 unsigned char *begalt;
1651 /* Place in the uncompiled pattern (i.e., the {) to
1652 which to go back if the interval is invalid. */
1653 const char *beg_interval;
1655 /* Address of the place where a forward jump should go to the end of
1656 the containing expression. Each alternative of an `or' -- except the
1657 last -- ends with a forward jump of this sort. */
1658 unsigned char *fixup_alt_jump = 0;
1660 /* Counts open-groups as they are encountered. Remembered for the
1661 matching close-group on the compile stack, so the same register
1662 number is put in the stop_memory as the start_memory. */
1663 regnum_t regnum = 0;
1666 DEBUG_PRINT1 ("\nCompiling pattern: ");
1669 unsigned debug_count;
1671 for (debug_count = 0; debug_count < size; debug_count++)
1672 putchar (pattern[debug_count]);
1677 /* Initialize the compile stack. */
1678 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1679 if (compile_stack.stack == NULL)
1682 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1683 compile_stack.avail = 0;
1685 /* Initialize the pattern buffer. */
1686 bufp->syntax = syntax;
1687 bufp->fastmap_accurate = 0;
1688 bufp->not_bol = bufp->not_eol = 0;
1690 /* Set `used' to zero, so that if we return an error, the pattern
1691 printer (for debugging) will think there's no pattern. We reset it
1695 /* Always count groups, whether or not bufp->no_sub is set. */
1698 #if !defined (emacs) && !defined (SYNTAX_TABLE)
1699 /* Initialize the syntax table. */
1700 init_syntax_once ();
1703 if (bufp->allocated == 0)
1706 { /* If zero allocated, but buffer is non-null, try to realloc
1707 enough space. This loses if buffer's address is bogus, but
1708 that is the user's responsibility. */
1709 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1712 { /* Caller did not allocate a buffer. Do it for them. */
1713 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1715 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1717 bufp->allocated = INIT_BUF_SIZE;
1720 begalt = b = bufp->buffer;
1722 /* Loop through the uncompiled pattern until we're at the end. */
1731 if ( /* If at start of pattern, it's an operator. */
1733 /* If context independent, it's an operator. */
1734 || syntax & RE_CONTEXT_INDEP_ANCHORS
1735 /* Otherwise, depends on what's come before. */
1736 || at_begline_loc_p (pattern, p, syntax))
1746 if ( /* If at end of pattern, it's an operator. */
1748 /* If context independent, it's an operator. */
1749 || syntax & RE_CONTEXT_INDEP_ANCHORS
1750 /* Otherwise, depends on what's next. */
1751 || at_endline_loc_p (p, pend, syntax))
1761 if ((syntax & RE_BK_PLUS_QM)
1762 || (syntax & RE_LIMITED_OPS))
1766 /* If there is no previous pattern... */
1769 if (syntax & RE_CONTEXT_INVALID_OPS)
1770 FREE_STACK_RETURN (REG_BADRPT);
1771 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
1776 /* Are we optimizing this jump? */
1777 boolean keep_string_p = false;
1779 /* 1 means zero (many) matches is allowed. */
1780 char zero_times_ok = 0, many_times_ok = 0;
1782 /* If there is a sequence of repetition chars, collapse it
1783 down to just one (the right one). We can't combine
1784 interval operators with these because of, e.g., `a{2}*',
1785 which should only match an even number of `a's. */
1789 zero_times_ok |= c != '+';
1790 many_times_ok |= c != '?';
1798 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
1801 else if (syntax & RE_BK_PLUS_QM && c == '\\')
1803 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1806 if (!(c1 == '+' || c1 == '?'))
1821 /* If we get here, we found another repeat character. */
1824 /* Star, etc. applied to an empty pattern is equivalent
1825 to an empty pattern. */
1829 /* Now we know whether or not zero matches is allowed
1830 and also whether or not two or more matches is allowed. */
1832 { /* More than one repetition is allowed, so put in at the
1833 end a backward relative jump from `b' to before the next
1834 jump we're going to put in below (which jumps from
1835 laststart to after this jump).
1837 But if we are at the `*' in the exact sequence `.*\n',
1838 insert an unconditional jump backwards to the .,
1839 instead of the beginning of the loop. This way we only
1840 push a failure point once, instead of every time
1841 through the loop. */
1842 assert (p - 1 > pattern);
1844 /* Allocate the space for the jump. */
1845 GET_BUFFER_SPACE (3);
1847 /* We know we are not at the first character of the pattern,
1848 because laststart was nonzero. And we've already
1849 incremented `p', by the way, to be the character after
1850 the `*'. Do we have to do something analogous here
1851 for null bytes, because of RE_DOT_NOT_NULL? */
1852 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
1854 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
1855 && !(syntax & RE_DOT_NEWLINE))
1856 { /* We have .*\n. */
1857 STORE_JUMP (jump, b, laststart);
1858 keep_string_p = true;
1861 /* Anything else. */
1862 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
1864 /* We've added more stuff to the buffer. */
1868 /* On failure, jump from laststart to b + 3, which will be the
1869 end of the buffer after this jump is inserted. */
1870 GET_BUFFER_SPACE (3);
1871 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
1879 /* At least one repetition is required, so insert a
1880 `dummy_failure_jump' before the initial
1881 `on_failure_jump' instruction of the loop. This
1882 effects a skip over that instruction the first time
1883 we hit that loop. */
1884 GET_BUFFER_SPACE (3);
1885 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
1900 boolean had_char_class = false;
1902 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1904 /* Ensure that we have enough space to push a charset: the
1905 opcode, the length count, and the bitset; 34 bytes in all. */
1906 GET_BUFFER_SPACE (34);
1910 /* We test `*p == '^' twice, instead of using an if
1911 statement, so we only need one BUF_PUSH. */
1912 BUF_PUSH (*p == '^' ? charset_not : charset);
1916 /* Remember the first position in the bracket expression. */
1919 /* Push the number of bytes in the bitmap. */
1920 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
1922 /* Clear the whole map. */
1923 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
1925 /* charset_not matches newline according to a syntax bit. */
1926 if ((re_opcode_t) b[-2] == charset_not
1927 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
1928 SET_LIST_BIT ('\n');
1930 /* Read in characters and ranges, setting map bits. */
1933 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1937 /* \ might escape characters inside [...] and [^...]. */
1938 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
1940 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
1947 /* Could be the end of the bracket expression. If it's
1948 not (i.e., when the bracket expression is `[]' so
1949 far), the ']' character bit gets set way below. */
1950 if (c == ']' && p != p1 + 1)
1953 /* Look ahead to see if it's a range when the last thing
1954 was a character class. */
1955 if (had_char_class && c == '-' && *p != ']')
1956 FREE_STACK_RETURN (REG_ERANGE);
1958 /* Look ahead to see if it's a range when the last thing
1959 was a character: if this is a hyphen not at the
1960 beginning or the end of a list, then it's the range
1963 && !(p - 2 >= pattern && p[-2] == '[')
1964 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
1968 = compile_range (&p, pend, translate, syntax, b);
1969 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
1972 else if (p[0] == '-' && p[1] != ']')
1973 { /* This handles ranges made up of characters only. */
1976 /* Move past the `-'. */
1979 ret = compile_range (&p, pend, translate, syntax, b);
1980 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
1983 /* See if we're at the beginning of a possible character
1986 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
1987 { /* Leave room for the null. */
1988 char str[CHAR_CLASS_MAX_LENGTH + 1];
1993 /* If pattern is `[[:'. */
1994 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
1999 if (c == ':' || c == ']' || p == pend
2000 || c1 == CHAR_CLASS_MAX_LENGTH)
2006 /* If isn't a word bracketed by `[:' and:`]':
2007 undo the ending character, the letters, and leave
2008 the leading `:' and `[' (but set bits for them). */
2009 if (c == ':' && *p == ']')
2012 boolean is_alnum = STREQ (str, "alnum");
2013 boolean is_alpha = STREQ (str, "alpha");
2014 boolean is_blank = STREQ (str, "blank");
2015 boolean is_cntrl = STREQ (str, "cntrl");
2016 boolean is_digit = STREQ (str, "digit");
2017 boolean is_graph = STREQ (str, "graph");
2018 boolean is_lower = STREQ (str, "lower");
2019 boolean is_print = STREQ (str, "print");
2020 boolean is_punct = STREQ (str, "punct");
2021 boolean is_space = STREQ (str, "space");
2022 boolean is_upper = STREQ (str, "upper");
2023 boolean is_xdigit = STREQ (str, "xdigit");
2025 if (!IS_CHAR_CLASS (str))
2026 FREE_STACK_RETURN (REG_ECTYPE);
2028 /* Throw away the ] at the end of the character
2032 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2034 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2036 /* This was split into 3 if's to
2037 avoid an arbitrary limit in some compiler. */
2038 if ( (is_alnum && ISALNUM (ch))
2039 || (is_alpha && ISALPHA (ch))
2040 || (is_blank && ISBLANK (ch))
2041 || (is_cntrl && ISCNTRL (ch)))
2043 if ( (is_digit && ISDIGIT (ch))
2044 || (is_graph && ISGRAPH (ch))
2045 || (is_lower && ISLOWER (ch))
2046 || (is_print && ISPRINT (ch)))
2048 if ( (is_punct && ISPUNCT (ch))
2049 || (is_space && ISSPACE (ch))
2050 || (is_upper && ISUPPER (ch))
2051 || (is_xdigit && ISXDIGIT (ch)))
2054 had_char_class = true;
2063 had_char_class = false;
2068 had_char_class = false;
2073 /* Discard any (non)matching list bytes that are all 0 at the
2074 end of the map. Decrease the map-length byte too. */
2075 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2083 if (syntax & RE_NO_BK_PARENS)
2090 if (syntax & RE_NO_BK_PARENS)
2097 if (syntax & RE_NEWLINE_ALT)
2104 if (syntax & RE_NO_BK_VBAR)
2111 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2112 goto handle_interval;
2118 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2120 /* Do not translate the character after the \, so that we can
2121 distinguish, e.g., \B from \b, even if we normally would
2122 translate, e.g., B to b. */
2128 if (syntax & RE_NO_BK_PARENS)
2129 goto normal_backslash;
2135 if (COMPILE_STACK_FULL)
2137 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2138 compile_stack_elt_t);
2139 if (compile_stack.stack == NULL) return REG_ESPACE;
2141 compile_stack.size <<= 1;
2144 /* These are the values to restore when we hit end of this
2145 group. They are all relative offsets, so that if the
2146 whole pattern moves because of realloc, they will still
2148 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2149 COMPILE_STACK_TOP.fixup_alt_jump
2150 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2151 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2152 COMPILE_STACK_TOP.regnum = regnum;
2154 /* We will eventually replace the 0 with the number of
2155 groups inner to this one. But do not push a
2156 start_memory for groups beyond the last one we can
2157 represent in the compiled pattern. */
2158 if (regnum <= MAX_REGNUM)
2160 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2161 BUF_PUSH_3 (start_memory, regnum, 0);
2164 compile_stack.avail++;
2169 /* If we've reached MAX_REGNUM groups, then this open
2170 won't actually generate any code, so we'll have to
2171 clear pending_exact explicitly. */
2177 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2179 if (COMPILE_STACK_EMPTY)
2180 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2181 goto normal_backslash;
2183 FREE_STACK_RETURN (REG_ERPAREN);
2187 { /* Push a dummy failure point at the end of the
2188 alternative for a possible future
2189 `pop_failure_jump' to pop. See comments at
2190 `push_dummy_failure' in `re_match_2'. */
2191 BUF_PUSH (push_dummy_failure);
2193 /* We allocated space for this jump when we assigned
2194 to `fixup_alt_jump', in the `handle_alt' case below. */
2195 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2198 /* See similar code for backslashed left paren above. */
2199 if (COMPILE_STACK_EMPTY)
2200 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2203 FREE_STACK_RETURN (REG_ERPAREN);
2205 /* Since we just checked for an empty stack above, this
2206 ``can't happen''. */
2207 assert (compile_stack.avail != 0);
2209 /* We don't just want to restore into `regnum', because
2210 later groups should continue to be numbered higher,
2211 as in `(ab)c(de)' -- the second group is #2. */
2212 regnum_t this_group_regnum;
2214 compile_stack.avail--;
2215 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2217 = COMPILE_STACK_TOP.fixup_alt_jump
2218 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2220 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2221 this_group_regnum = COMPILE_STACK_TOP.regnum;
2222 /* If we've reached MAX_REGNUM groups, then this open
2223 won't actually generate any code, so we'll have to
2224 clear pending_exact explicitly. */
2227 /* We're at the end of the group, so now we know how many
2228 groups were inside this one. */
2229 if (this_group_regnum <= MAX_REGNUM)
2231 unsigned char *inner_group_loc
2232 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2234 *inner_group_loc = regnum - this_group_regnum;
2235 BUF_PUSH_3 (stop_memory, this_group_regnum,
2236 regnum - this_group_regnum);
2242 case '|': /* `\|'. */
2243 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2244 goto normal_backslash;
2246 if (syntax & RE_LIMITED_OPS)
2249 /* Insert before the previous alternative a jump which
2250 jumps to this alternative if the former fails. */
2251 GET_BUFFER_SPACE (3);
2252 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2256 /* The alternative before this one has a jump after it
2257 which gets executed if it gets matched. Adjust that
2258 jump so it will jump to this alternative's analogous
2259 jump (put in below, which in turn will jump to the next
2260 (if any) alternative's such jump, etc.). The last such
2261 jump jumps to the correct final destination. A picture:
2267 If we are at `b', then fixup_alt_jump right now points to a
2268 three-byte space after `a'. We'll put in the jump, set
2269 fixup_alt_jump to right after `b', and leave behind three
2270 bytes which we'll fill in when we get to after `c'. */
2273 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2275 /* Mark and leave space for a jump after this alternative,
2276 to be filled in later either by next alternative or
2277 when know we're at the end of a series of alternatives. */
2279 GET_BUFFER_SPACE (3);
2288 /* If \{ is a literal. */
2289 if (!(syntax & RE_INTERVALS)
2290 /* If we're at `\{' and it's not the open-interval
2292 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2293 || (p - 2 == pattern && p == pend))
2294 goto normal_backslash;
2298 /* If got here, then the syntax allows intervals. */
2300 /* At least (most) this many matches must be made. */
2301 int lower_bound = -1, upper_bound = -1;
2303 beg_interval = p - 1;
2307 if (syntax & RE_NO_BK_BRACES)
2308 goto unfetch_interval;
2310 FREE_STACK_RETURN (REG_EBRACE);
2313 GET_UNSIGNED_NUMBER (lower_bound);
2317 GET_UNSIGNED_NUMBER (upper_bound);
2318 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2321 /* Interval such as `{1}' => match exactly once. */
2322 upper_bound = lower_bound;
2324 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2325 || lower_bound > upper_bound)
2327 if (syntax & RE_NO_BK_BRACES)
2328 goto unfetch_interval;
2330 FREE_STACK_RETURN (REG_BADBR);
2333 if (!(syntax & RE_NO_BK_BRACES))
2335 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2342 if (syntax & RE_NO_BK_BRACES)
2343 goto unfetch_interval;
2345 FREE_STACK_RETURN (REG_BADBR);
2348 /* We just parsed a valid interval. */
2350 /* If it's invalid to have no preceding re. */
2353 if (syntax & RE_CONTEXT_INVALID_OPS)
2354 FREE_STACK_RETURN (REG_BADRPT);
2355 else if (syntax & RE_CONTEXT_INDEP_OPS)
2358 goto unfetch_interval;
2361 /* If the upper bound is zero, don't want to succeed at
2362 all; jump from `laststart' to `b + 3', which will be
2363 the end of the buffer after we insert the jump. */
2364 if (upper_bound == 0)
2366 GET_BUFFER_SPACE (3);
2367 INSERT_JUMP (jump, laststart, b + 3);
2371 /* Otherwise, we have a nontrivial interval. When
2372 we're all done, the pattern will look like:
2373 set_number_at <jump count> <upper bound>
2374 set_number_at <succeed_n count> <lower bound>
2375 succeed_n <after jump addr> <succeed_n count>
2377 jump_n <succeed_n addr> <jump count>
2378 (The upper bound and `jump_n' are omitted if
2379 `upper_bound' is 1, though.) */
2381 { /* If the upper bound is > 1, we need to insert
2382 more at the end of the loop. */
2383 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2385 GET_BUFFER_SPACE (nbytes);
2387 /* Initialize lower bound of the `succeed_n', even
2388 though it will be set during matching by its
2389 attendant `set_number_at' (inserted next),
2390 because `re_compile_fastmap' needs to know.
2391 Jump to the `jump_n' we might insert below. */
2392 INSERT_JUMP2 (succeed_n, laststart,
2393 b + 5 + (upper_bound > 1) * 5,
2397 /* Code to initialize the lower bound. Insert
2398 before the `succeed_n'. The `5' is the last two
2399 bytes of this `set_number_at', plus 3 bytes of
2400 the following `succeed_n'. */
2401 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2404 if (upper_bound > 1)
2405 { /* More than one repetition is allowed, so
2406 append a backward jump to the `succeed_n'
2407 that starts this interval.
2409 When we've reached this during matching,
2410 we'll have matched the interval once, so
2411 jump back only `upper_bound - 1' times. */
2412 STORE_JUMP2 (jump_n, b, laststart + 5,
2416 /* The location we want to set is the second
2417 parameter of the `jump_n'; that is `b-2' as
2418 an absolute address. `laststart' will be
2419 the `set_number_at' we're about to insert;
2420 `laststart+3' the number to set, the source
2421 for the relative address. But we are
2422 inserting into the middle of the pattern --
2423 so everything is getting moved up by 5.
2424 Conclusion: (b - 2) - (laststart + 3) + 5,
2425 i.e., b - laststart.
2427 We insert this at the beginning of the loop
2428 so that if we fail during matching, we'll
2429 reinitialize the bounds. */
2430 insert_op2 (set_number_at, laststart, b - laststart,
2431 upper_bound - 1, b);
2436 beg_interval = NULL;
2441 /* If an invalid interval, match the characters as literals. */
2442 assert (beg_interval);
2444 beg_interval = NULL;
2446 /* normal_char and normal_backslash need `c'. */
2449 if (!(syntax & RE_NO_BK_BRACES))
2451 if (p > pattern && p[-1] == '\\')
2452 goto normal_backslash;
2457 /* There is no way to specify the before_dot and after_dot
2458 operators. rms says this is ok. --karl */
2466 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2472 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2479 BUF_PUSH (wordchar);
2485 BUF_PUSH (notwordchar);
2498 BUF_PUSH (wordbound);
2502 BUF_PUSH (notwordbound);
2513 case '1': case '2': case '3': case '4': case '5':
2514 case '6': case '7': case '8': case '9':
2515 if (syntax & RE_NO_BK_REFS)
2521 FREE_STACK_RETURN (REG_ESUBREG);
2523 /* Can't back reference to a subexpression if inside of it. */
2524 if (group_in_compile_stack (compile_stack, c1))
2528 BUF_PUSH_2 (duplicate, c1);
2534 if (syntax & RE_BK_PLUS_QM)
2537 goto normal_backslash;
2541 /* You might think it would be useful for \ to mean
2542 not to translate; but if we don't translate it
2543 it will never match anything. */
2551 /* Expects the character in `c'. */
2553 /* If no exactn currently being built. */
2556 /* If last exactn not at current position. */
2557 || pending_exact + *pending_exact + 1 != b
2559 /* We have only one byte following the exactn for the count. */
2560 || *pending_exact == (1 << BYTEWIDTH) - 1
2562 /* If followed by a repetition operator. */
2563 || *p == '*' || *p == '^'
2564 || ((syntax & RE_BK_PLUS_QM)
2565 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2566 : (*p == '+' || *p == '?'))
2567 || ((syntax & RE_INTERVALS)
2568 && ((syntax & RE_NO_BK_BRACES)
2570 : (p[0] == '\\' && p[1] == '{'))))
2572 /* Start building a new exactn. */
2576 BUF_PUSH_2 (exactn, 0);
2577 pending_exact = b - 1;
2584 } /* while p != pend */
2587 /* Through the pattern now. */
2590 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2592 if (!COMPILE_STACK_EMPTY)
2593 FREE_STACK_RETURN (REG_EPAREN);
2595 /* If we don't want backtracking, force success
2596 the first time we reach the end of the compiled pattern. */
2597 if (syntax & RE_NO_POSIX_BACKTRACKING)
2600 free (compile_stack.stack);
2602 /* We have succeeded; set the length of the buffer. */
2603 bufp->used = b - bufp->buffer;
2608 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2609 print_compiled_pattern (bufp);
2613 #ifndef MATCH_MAY_ALLOCATE
2614 /* Initialize the failure stack to the largest possible stack. This
2615 isn't necessary unless we're trying to avoid calling alloca in
2616 the search and match routines. */
2618 int num_regs = bufp->re_nsub + 1;
2620 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2621 is strictly greater than re_max_failures, the largest possible stack
2622 is 2 * re_max_failures failure points. */
2623 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2625 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2628 if (! fail_stack.stack)
2630 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2631 * sizeof (fail_stack_elt_t));
2634 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2636 * sizeof (fail_stack_elt_t)));
2637 #else /* not emacs */
2638 if (! fail_stack.stack)
2640 = (fail_stack_elt_t *) malloc (fail_stack.size
2641 * sizeof (fail_stack_elt_t));
2644 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2646 * sizeof (fail_stack_elt_t)));
2647 #endif /* not emacs */
2650 regex_grow_registers (num_regs);
2652 #endif /* not MATCH_MAY_ALLOCATE */
2655 } /* regex_compile */
2657 /* Subroutines for `regex_compile'. */
2659 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2662 store_op1 (op, loc, arg)
2667 *loc = (unsigned char) op;
2668 STORE_NUMBER (loc + 1, arg);
2672 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2675 store_op2 (op, loc, arg1, arg2)
2680 *loc = (unsigned char) op;
2681 STORE_NUMBER (loc + 1, arg1);
2682 STORE_NUMBER (loc + 3, arg2);
2686 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
2687 for OP followed by two-byte integer parameter ARG. */
2690 insert_op1 (op, loc, arg, end)
2696 register unsigned char *pfrom = end;
2697 register unsigned char *pto = end + 3;
2699 while (pfrom != loc)
2702 store_op1 (op, loc, arg);
2706 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
2709 insert_op2 (op, loc, arg1, arg2, end)
2715 register unsigned char *pfrom = end;
2716 register unsigned char *pto = end + 5;
2718 while (pfrom != loc)
2721 store_op2 (op, loc, arg1, arg2);
2725 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
2726 after an alternative or a begin-subexpression. We assume there is at
2727 least one character before the ^. */
2730 at_begline_loc_p (pattern, p, syntax)
2731 const char *pattern, *p;
2732 reg_syntax_t syntax;
2734 const char *prev = p - 2;
2735 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
2738 /* After a subexpression? */
2739 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
2740 /* After an alternative? */
2741 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
2745 /* The dual of at_begline_loc_p. This one is for $. We assume there is
2746 at least one character after the $, i.e., `P < PEND'. */
2749 at_endline_loc_p (p, pend, syntax)
2750 const char *p, *pend;
2753 const char *next = p;
2754 boolean next_backslash = *next == '\\';
2755 const char *next_next = p + 1 < pend ? p + 1 : NULL;
2758 /* Before a subexpression? */
2759 (syntax & RE_NO_BK_PARENS ? *next == ')'
2760 : next_backslash && next_next && *next_next == ')')
2761 /* Before an alternative? */
2762 || (syntax & RE_NO_BK_VBAR ? *next == '|'
2763 : next_backslash && next_next && *next_next == '|');
2767 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
2768 false if it's not. */
2771 group_in_compile_stack (compile_stack, regnum)
2772 compile_stack_type compile_stack;
2777 for (this_element = compile_stack.avail - 1;
2780 if (compile_stack.stack[this_element].regnum == regnum)
2787 /* Read the ending character of a range (in a bracket expression) from the
2788 uncompiled pattern *P_PTR (which ends at PEND). We assume the
2789 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
2790 Then we set the translation of all bits between the starting and
2791 ending characters (inclusive) in the compiled pattern B.
2793 Return an error code.
2795 We use these short variable names so we can use the same macros as
2796 `regex_compile' itself. */
2798 static reg_errcode_t
2799 compile_range (p_ptr, pend, translate, syntax, b)
2800 const char **p_ptr, *pend;
2802 reg_syntax_t syntax;
2807 const char *p = *p_ptr;
2808 int range_start, range_end;
2813 /* Even though the pattern is a signed `char *', we need to fetch
2814 with unsigned char *'s; if the high bit of the pattern character
2815 is set, the range endpoints will be negative if we fetch using a
2818 We also want to fetch the endpoints without translating them; the
2819 appropriate translation is done in the bit-setting loop below. */
2820 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
2821 range_start = ((const unsigned char *) p)[-2];
2822 range_end = ((const unsigned char *) p)[0];
2824 /* Have to increment the pointer into the pattern string, so the
2825 caller isn't still at the ending character. */
2828 /* If the start is after the end, the range is empty. */
2829 if (range_start > range_end)
2830 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
2832 /* Here we see why `this_char' has to be larger than an `unsigned
2833 char' -- the range is inclusive, so if `range_end' == 0xff
2834 (assuming 8-bit characters), we would otherwise go into an infinite
2835 loop, since all characters <= 0xff. */
2836 for (this_char = range_start; this_char <= range_end; this_char++)
2838 SET_LIST_BIT (TRANSLATE (this_char));
2844 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
2845 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
2846 characters can start a string that matches the pattern. This fastmap
2847 is used by re_search to skip quickly over impossible starting points.
2849 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
2850 area as BUFP->fastmap.
2852 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
2855 Returns 0 if we succeed, -2 if an internal error. */
2858 re_compile_fastmap (bufp)
2859 struct re_pattern_buffer *bufp;
2862 #ifdef MATCH_MAY_ALLOCATE
2863 fail_stack_type fail_stack;
2865 #ifndef REGEX_MALLOC
2868 /* We don't push any register information onto the failure stack. */
2869 unsigned num_regs = 0;
2871 register char *fastmap = bufp->fastmap;
2872 unsigned char *pattern = bufp->buffer;
2873 unsigned long size = bufp->used;
2874 unsigned char *p = pattern;
2875 register unsigned char *pend = pattern + size;
2877 /* This holds the pointer to the failure stack, when
2878 it is allocated relocatably. */
2879 fail_stack_elt_t *failure_stack_ptr;
2881 /* Assume that each path through the pattern can be null until
2882 proven otherwise. We set this false at the bottom of switch
2883 statement, to which we get only if a particular path doesn't
2884 match the empty string. */
2885 boolean path_can_be_null = true;
2887 /* We aren't doing a `succeed_n' to begin with. */
2888 boolean succeed_n_p = false;
2890 assert (fastmap != NULL && p != NULL);
2893 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
2894 bufp->fastmap_accurate = 1; /* It will be when we're done. */
2895 bufp->can_be_null = 0;
2899 if (p == pend || *p == succeed)
2901 /* We have reached the (effective) end of pattern. */
2902 if (!FAIL_STACK_EMPTY ())
2904 bufp->can_be_null |= path_can_be_null;
2906 /* Reset for next path. */
2907 path_can_be_null = true;
2909 p = fail_stack.stack[--fail_stack.avail].pointer;
2917 /* We should never be about to go beyond the end of the pattern. */
2920 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
2923 /* I guess the idea here is to simply not bother with a fastmap
2924 if a backreference is used, since it's too hard to figure out
2925 the fastmap for the corresponding group. Setting
2926 `can_be_null' stops `re_search_2' from using the fastmap, so
2927 that is all we do. */
2929 bufp->can_be_null = 1;
2933 /* Following are the cases which match a character. These end
2942 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
2943 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
2949 /* Chars beyond end of map must be allowed. */
2950 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
2953 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
2954 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
2960 for (j = 0; j < (1 << BYTEWIDTH); j++)
2961 if (SYNTAX (j) == Sword)
2967 for (j = 0; j < (1 << BYTEWIDTH); j++)
2968 if (SYNTAX (j) != Sword)
2975 int fastmap_newline = fastmap['\n'];
2977 /* `.' matches anything ... */
2978 for (j = 0; j < (1 << BYTEWIDTH); j++)
2981 /* ... except perhaps newline. */
2982 if (!(bufp->syntax & RE_DOT_NEWLINE))
2983 fastmap['\n'] = fastmap_newline;
2985 /* Return if we have already set `can_be_null'; if we have,
2986 then the fastmap is irrelevant. Something's wrong here. */
2987 else if (bufp->can_be_null)
2990 /* Otherwise, have to check alternative paths. */
2997 for (j = 0; j < (1 << BYTEWIDTH); j++)
2998 if (SYNTAX (j) == (enum syntaxcode) k)
3005 for (j = 0; j < (1 << BYTEWIDTH); j++)
3006 if (SYNTAX (j) != (enum syntaxcode) k)
3011 /* All cases after this match the empty string. These end with
3019 #endif /* not emacs */
3031 case push_dummy_failure:
3036 case pop_failure_jump:
3037 case maybe_pop_jump:
3040 case dummy_failure_jump:
3041 EXTRACT_NUMBER_AND_INCR (j, p);
3046 /* Jump backward implies we just went through the body of a
3047 loop and matched nothing. Opcode jumped to should be
3048 `on_failure_jump' or `succeed_n'. Just treat it like an
3049 ordinary jump. For a * loop, it has pushed its failure
3050 point already; if so, discard that as redundant. */
3051 if ((re_opcode_t) *p != on_failure_jump
3052 && (re_opcode_t) *p != succeed_n)
3056 EXTRACT_NUMBER_AND_INCR (j, p);
3059 /* If what's on the stack is where we are now, pop it. */
3060 if (!FAIL_STACK_EMPTY ()
3061 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3067 case on_failure_jump:
3068 case on_failure_keep_string_jump:
3069 handle_on_failure_jump:
3070 EXTRACT_NUMBER_AND_INCR (j, p);
3072 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3073 end of the pattern. We don't want to push such a point,
3074 since when we restore it above, entering the switch will
3075 increment `p' past the end of the pattern. We don't need
3076 to push such a point since we obviously won't find any more
3077 fastmap entries beyond `pend'. Such a pattern can match
3078 the null string, though. */
3081 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3083 REGEX_FREE_STACK (fail_stack.stack);
3088 bufp->can_be_null = 1;
3092 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3093 succeed_n_p = false;
3100 /* Get to the number of times to succeed. */
3103 /* Increment p past the n for when k != 0. */
3104 EXTRACT_NUMBER_AND_INCR (k, p);
3108 succeed_n_p = true; /* Spaghetti code alert. */
3109 goto handle_on_failure_jump;
3126 abort (); /* We have listed all the cases. */
3129 /* Getting here means we have found the possible starting
3130 characters for one path of the pattern -- and that the empty
3131 string does not match. We need not follow this path further.
3132 Instead, look at the next alternative (remembered on the
3133 stack), or quit if no more. The test at the top of the loop
3134 does these things. */
3135 path_can_be_null = false;
3139 /* Set `can_be_null' for the last path (also the first path, if the
3140 pattern is empty). */
3141 bufp->can_be_null |= path_can_be_null;
3144 if (!FAIL_STACK_EMPTY ())
3145 REGEX_FREE_STACK (fail_stack.stack);
3147 } /* re_compile_fastmap */
3149 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3150 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3151 this memory for recording register information. STARTS and ENDS
3152 must be allocated using the malloc library routine, and must each
3153 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3155 If NUM_REGS == 0, then subsequent matches should allocate their own
3158 Unless this function is called, the first search or match using
3159 PATTERN_BUFFER will allocate its own register data, without
3160 freeing the old data. */
3163 re_set_registers (bufp, regs, num_regs, starts, ends)
3164 struct re_pattern_buffer *bufp;
3165 struct re_registers *regs;
3167 regoff_t *starts, *ends;
3171 bufp->regs_allocated = REGS_REALLOCATE;
3172 regs->num_regs = num_regs;
3173 regs->start = starts;
3178 bufp->regs_allocated = REGS_UNALLOCATED;
3180 regs->start = regs->end = (regoff_t *) 0;
3184 /* Searching routines. */
3186 /* Like re_search_2, below, but only one string is specified, and
3187 doesn't let you say where to stop matching. */
3190 re_search (bufp, string, size, startpos, range, regs)
3191 struct re_pattern_buffer *bufp;
3193 int size, startpos, range;
3194 struct re_registers *regs;
3196 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3201 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3202 virtual concatenation of STRING1 and STRING2, starting first at index
3203 STARTPOS, then at STARTPOS + 1, and so on.
3205 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3207 RANGE is how far to scan while trying to match. RANGE = 0 means try
3208 only at STARTPOS; in general, the last start tried is STARTPOS +
3211 In REGS, return the indices of the virtual concatenation of STRING1
3212 and STRING2 that matched the entire BUFP->buffer and its contained
3215 Do not consider matching one past the index STOP in the virtual
3216 concatenation of STRING1 and STRING2.
3218 We return either the position in the strings at which the match was
3219 found, -1 if no match, or -2 if error (such as failure
3223 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3224 struct re_pattern_buffer *bufp;
3225 const char *string1, *string2;
3229 struct re_registers *regs;
3233 register char *fastmap = bufp->fastmap;
3234 register char *translate = bufp->translate;
3235 int total_size = size1 + size2;
3236 int endpos = startpos + range;
3238 /* Check for out-of-range STARTPOS. */
3239 if (startpos < 0 || startpos > total_size)
3242 /* Fix up RANGE if it might eventually take us outside
3243 the virtual concatenation of STRING1 and STRING2. */
3245 range = -1 - startpos;
3246 else if (endpos > total_size)
3247 range = total_size - startpos;
3249 /* If the search isn't to be a backwards one, don't waste time in a
3250 search for a pattern that must be anchored. */
3251 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3259 /* Update the fastmap now if not correct already. */
3260 if (fastmap && !bufp->fastmap_accurate)
3261 if (re_compile_fastmap (bufp) == -2)
3264 /* Loop through the string, looking for a place to start matching. */
3267 /* If a fastmap is supplied, skip quickly over characters that
3268 cannot be the start of a match. If the pattern can match the
3269 null string, however, we don't need to skip characters; we want
3270 the first null string. */
3271 if (fastmap && startpos < total_size && !bufp->can_be_null)
3273 if (range > 0) /* Searching forwards. */
3275 register const char *d;
3276 register int lim = 0;
3279 if (startpos < size1 && startpos + range >= size1)
3280 lim = range - (size1 - startpos);
3282 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
3284 /* Written out as an if-else to avoid testing `translate'
3288 && !fastmap[(unsigned char)
3289 translate[(unsigned char) *d++]])
3292 while (range > lim && !fastmap[(unsigned char) *d++])
3295 startpos += irange - range;
3297 else /* Searching backwards. */
3299 register char c = (size1 == 0 || startpos >= size1
3300 ? string2[startpos - size1]
3301 : string1[startpos]);
3303 if (!fastmap[(unsigned char) TRANSLATE (c)])
3308 /* If can't match the null string, and that's all we have left, fail. */
3309 if (range >= 0 && startpos == total_size && fastmap
3310 && !bufp->can_be_null)
3313 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3314 startpos, regs, stop);
3315 #ifndef REGEX_MALLOC
3344 /* Declarations and macros for re_match_2. */
3346 static int bcmp_translate ();
3347 static boolean alt_match_null_string_p (),
3348 common_op_match_null_string_p (),
3349 group_match_null_string_p ();
3351 /* This converts PTR, a pointer into one of the search strings `string1'
3352 and `string2' into an offset from the beginning of that string. */
3353 #define POINTER_TO_OFFSET(ptr) \
3354 (FIRST_STRING_P (ptr) \
3355 ? ((regoff_t) ((ptr) - string1)) \
3356 : ((regoff_t) ((ptr) - string2 + size1)))
3358 /* Macros for dealing with the split strings in re_match_2. */
3360 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3362 /* Call before fetching a character with *d. This switches over to
3363 string2 if necessary. */
3364 #define PREFETCH() \
3367 /* End of string2 => fail. */ \
3368 if (dend == end_match_2) \
3370 /* End of string1 => advance to string2. */ \
3372 dend = end_match_2; \
3376 /* Test if at very beginning or at very end of the virtual concatenation
3377 of `string1' and `string2'. If only one string, it's `string2'. */
3378 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3379 #define AT_STRINGS_END(d) ((d) == end2)
3382 /* Test if D points to a character which is word-constituent. We have
3383 two special cases to check for: if past the end of string1, look at
3384 the first character in string2; and if before the beginning of
3385 string2, look at the last character in string1. */
3386 #define WORDCHAR_P(d) \
3387 (SYNTAX ((d) == end1 ? *string2 \
3388 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3391 /* Test if the character before D and the one at D differ with respect
3392 to being word-constituent. */
3393 #define AT_WORD_BOUNDARY(d) \
3394 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3395 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3398 /* Free everything we malloc. */
3399 #ifdef MATCH_MAY_ALLOCATE
3400 #define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
3401 #define FREE_VARIABLES() \
3403 REGEX_FREE_STACK (fail_stack.stack); \
3404 FREE_VAR (regstart); \
3405 FREE_VAR (regend); \
3406 FREE_VAR (old_regstart); \
3407 FREE_VAR (old_regend); \
3408 FREE_VAR (best_regstart); \
3409 FREE_VAR (best_regend); \
3410 FREE_VAR (reg_info); \
3411 FREE_VAR (reg_dummy); \
3412 FREE_VAR (reg_info_dummy); \
3415 #define FREE_VARIABLES() /* Do nothing! */
3416 #endif /* not MATCH_MAY_ALLOCATE */
3418 /* These values must meet several constraints. They must not be valid
3419 register values; since we have a limit of 255 registers (because
3420 we use only one byte in the pattern for the register number), we can
3421 use numbers larger than 255. They must differ by 1, because of
3422 NUM_FAILURE_ITEMS above. And the value for the lowest register must
3423 be larger than the value for the highest register, so we do not try
3424 to actually save any registers when none are active. */
3425 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
3426 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
3428 /* Matching routines. */
3430 #ifndef emacs /* Emacs never uses this. */
3431 /* re_match is like re_match_2 except it takes only a single string. */
3434 re_match (bufp, string, size, pos, regs)
3435 struct re_pattern_buffer *bufp;
3438 struct re_registers *regs;
3440 int result = re_match_2_internal (bufp, NULL, 0, string, size,
3445 #endif /* not emacs */
3448 /* re_match_2 matches the compiled pattern in BUFP against the
3449 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
3450 and SIZE2, respectively). We start matching at POS, and stop
3453 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
3454 store offsets for the substring each group matched in REGS. See the
3455 documentation for exactly how many groups we fill.
3457 We return -1 if no match, -2 if an internal error (such as the
3458 failure stack overflowing). Otherwise, we return the length of the
3459 matched substring. */
3462 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
3463 struct re_pattern_buffer *bufp;
3464 const char *string1, *string2;
3467 struct re_registers *regs;
3470 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
3476 /* This is a separate function so that we can force an alloca cleanup
3479 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
3480 struct re_pattern_buffer *bufp;
3481 const char *string1, *string2;
3484 struct re_registers *regs;
3487 /* General temporaries. */
3491 /* Just past the end of the corresponding string. */
3492 const char *end1, *end2;
3494 /* Pointers into string1 and string2, just past the last characters in
3495 each to consider matching. */
3496 const char *end_match_1, *end_match_2;
3498 /* Where we are in the data, and the end of the current string. */
3499 const char *d, *dend;
3501 /* Where we are in the pattern, and the end of the pattern. */
3502 unsigned char *p = bufp->buffer;
3503 register unsigned char *pend = p + bufp->used;
3505 /* Mark the opcode just after a start_memory, so we can test for an
3506 empty subpattern when we get to the stop_memory. */
3507 unsigned char *just_past_start_mem = 0;
3509 /* We use this to map every character in the string. */
3510 char *translate = bufp->translate;
3512 /* Failure point stack. Each place that can handle a failure further
3513 down the line pushes a failure point on this stack. It consists of
3514 restart, regend, and reg_info for all registers corresponding to
3515 the subexpressions we're currently inside, plus the number of such
3516 registers, and, finally, two char *'s. The first char * is where
3517 to resume scanning the pattern; the second one is where to resume
3518 scanning the strings. If the latter is zero, the failure point is
3519 a ``dummy''; if a failure happens and the failure point is a dummy,
3520 it gets discarded and the next next one is tried. */
3521 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3522 fail_stack_type fail_stack;
3525 static unsigned failure_id = 0;
3526 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
3529 /* This holds the pointer to the failure stack, when
3530 it is allocated relocatably. */
3531 fail_stack_elt_t *failure_stack_ptr;
3533 /* We fill all the registers internally, independent of what we
3534 return, for use in backreferences. The number here includes
3535 an element for register zero. */
3536 unsigned num_regs = bufp->re_nsub + 1;
3538 /* The currently active registers. */
3539 unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG;
3540 unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG;
3542 /* Information on the contents of registers. These are pointers into
3543 the input strings; they record just what was matched (on this
3544 attempt) by a subexpression part of the pattern, that is, the
3545 regnum-th regstart pointer points to where in the pattern we began
3546 matching and the regnum-th regend points to right after where we
3547 stopped matching the regnum-th subexpression. (The zeroth register
3548 keeps track of what the whole pattern matches.) */
3549 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3550 const char **regstart, **regend;
3553 /* If a group that's operated upon by a repetition operator fails to
3554 match anything, then the register for its start will need to be
3555 restored because it will have been set to wherever in the string we
3556 are when we last see its open-group operator. Similarly for a
3558 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3559 const char **old_regstart, **old_regend;
3562 /* The is_active field of reg_info helps us keep track of which (possibly
3563 nested) subexpressions we are currently in. The matched_something
3564 field of reg_info[reg_num] helps us tell whether or not we have
3565 matched any of the pattern so far this time through the reg_num-th
3566 subexpression. These two fields get reset each time through any
3567 loop their register is in. */
3568 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
3569 register_info_type *reg_info;
3572 /* The following record the register info as found in the above
3573 variables when we find a match better than any we've seen before.
3574 This happens as we backtrack through the failure points, which in
3575 turn happens only if we have not yet matched the entire string. */
3576 unsigned best_regs_set = false;
3577 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3578 const char **best_regstart, **best_regend;
3581 /* Logically, this is `best_regend[0]'. But we don't want to have to
3582 allocate space for that if we're not allocating space for anything
3583 else (see below). Also, we never need info about register 0 for
3584 any of the other register vectors, and it seems rather a kludge to
3585 treat `best_regend' differently than the rest. So we keep track of
3586 the end of the best match so far in a separate variable. We
3587 initialize this to NULL so that when we backtrack the first time
3588 and need to test it, it's not garbage. */
3589 const char *match_end = NULL;
3591 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
3592 int set_regs_matched_done = 0;
3594 /* Used when we pop values we don't care about. */
3595 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
3596 const char **reg_dummy;
3597 register_info_type *reg_info_dummy;
3601 /* Counts the total number of registers pushed. */
3602 unsigned num_regs_pushed = 0;
3605 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
3609 #ifdef MATCH_MAY_ALLOCATE
3610 /* Do not bother to initialize all the register variables if there are
3611 no groups in the pattern, as it takes a fair amount of time. If
3612 there are groups, we include space for register 0 (the whole
3613 pattern), even though we never use it, since it simplifies the
3614 array indexing. We should fix this. */
3617 regstart = REGEX_TALLOC (num_regs, const char *);
3618 regend = REGEX_TALLOC (num_regs, const char *);
3619 old_regstart = REGEX_TALLOC (num_regs, const char *);
3620 old_regend = REGEX_TALLOC (num_regs, const char *);
3621 best_regstart = REGEX_TALLOC (num_regs, const char *);
3622 best_regend = REGEX_TALLOC (num_regs, const char *);
3623 reg_info = REGEX_TALLOC (num_regs, register_info_type);
3624 reg_dummy = REGEX_TALLOC (num_regs, const char *);
3625 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
3627 if (!(regstart && regend && old_regstart && old_regend && reg_info
3628 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
3636 /* We must initialize all our variables to NULL, so that
3637 `FREE_VARIABLES' doesn't try to free them. */
3638 regstart = regend = old_regstart = old_regend = best_regstart
3639 = best_regend = reg_dummy = NULL;
3640 reg_info = reg_info_dummy = (register_info_type *) NULL;
3642 #endif /* MATCH_MAY_ALLOCATE */
3644 /* The starting position is bogus. */
3645 if (pos < 0 || pos > size1 + size2)
3651 /* Initialize subexpression text positions to -1 to mark ones that no
3652 start_memory/stop_memory has been seen for. Also initialize the
3653 register information struct. */
3654 for (mcnt = 1; mcnt < num_regs; mcnt++)
3656 regstart[mcnt] = regend[mcnt]
3657 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
3659 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
3660 IS_ACTIVE (reg_info[mcnt]) = 0;
3661 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3662 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
3665 /* We move `string1' into `string2' if the latter's empty -- but not if
3666 `string1' is null. */
3667 if (size2 == 0 && string1 != NULL)
3674 end1 = string1 + size1;
3675 end2 = string2 + size2;
3677 /* Compute where to stop matching, within the two strings. */
3680 end_match_1 = string1 + stop;
3681 end_match_2 = string2;
3686 end_match_2 = string2 + stop - size1;
3689 /* `p' scans through the pattern as `d' scans through the data.
3690 `dend' is the end of the input string that `d' points within. `d'
3691 is advanced into the following input string whenever necessary, but
3692 this happens before fetching; therefore, at the beginning of the
3693 loop, `d' can be pointing at the end of a string, but it cannot
3695 if (size1 > 0 && pos <= size1)
3702 d = string2 + pos - size1;
3706 DEBUG_PRINT1 ("The compiled pattern is: ");
3707 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
3708 DEBUG_PRINT1 ("The string to match is: `");
3709 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
3710 DEBUG_PRINT1 ("'\n");
3712 /* This loops over pattern commands. It exits by returning from the
3713 function if the match is complete, or it drops through if the match
3714 fails at this starting point in the input data. */
3717 DEBUG_PRINT2 ("\n0x%x: ", p);
3720 { /* End of pattern means we might have succeeded. */
3721 DEBUG_PRINT1 ("end of pattern ... ");
3723 /* If we haven't matched the entire string, and we want the
3724 longest match, try backtracking. */
3725 if (d != end_match_2)
3727 /* 1 if this match ends in the same string (string1 or string2)
3728 as the best previous match. */
3729 boolean same_str_p = (FIRST_STRING_P (match_end)
3730 == MATCHING_IN_FIRST_STRING);
3731 /* 1 if this match is the best seen so far. */
3732 boolean best_match_p;
3734 /* AIX compiler got confused when this was combined
3735 with the previous declaration. */
3737 best_match_p = d > match_end;
3739 best_match_p = !MATCHING_IN_FIRST_STRING;
3741 DEBUG_PRINT1 ("backtracking.\n");
3743 if (!FAIL_STACK_EMPTY ())
3744 { /* More failure points to try. */
3746 /* If exceeds best match so far, save it. */
3747 if (!best_regs_set || best_match_p)
3749 best_regs_set = true;
3752 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
3754 for (mcnt = 1; mcnt < num_regs; mcnt++)
3756 best_regstart[mcnt] = regstart[mcnt];
3757 best_regend[mcnt] = regend[mcnt];
3763 /* If no failure points, don't restore garbage. And if
3764 last match is real best match, don't restore second
3766 else if (best_regs_set && !best_match_p)
3769 /* Restore best match. It may happen that `dend ==
3770 end_match_1' while the restored d is in string2.
3771 For example, the pattern `x.*y.*z' against the
3772 strings `x-' and `y-z-', if the two strings are
3773 not consecutive in memory. */
3774 DEBUG_PRINT1 ("Restoring best registers.\n");
3777 dend = ((d >= string1 && d <= end1)
3778 ? end_match_1 : end_match_2);
3780 for (mcnt = 1; mcnt < num_regs; mcnt++)
3782 regstart[mcnt] = best_regstart[mcnt];
3783 regend[mcnt] = best_regend[mcnt];
3786 } /* d != end_match_2 */
3789 DEBUG_PRINT1 ("Accepting match.\n");
3791 /* If caller wants register contents data back, do it. */
3792 if (regs && !bufp->no_sub)
3794 /* Have the register data arrays been allocated? */
3795 if (bufp->regs_allocated == REGS_UNALLOCATED)
3796 { /* No. So allocate them with malloc. We need one
3797 extra element beyond `num_regs' for the `-1' marker
3799 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
3800 regs->start = TALLOC (regs->num_regs, regoff_t);
3801 regs->end = TALLOC (regs->num_regs, regoff_t);
3802 if (regs->start == NULL || regs->end == NULL)
3807 bufp->regs_allocated = REGS_REALLOCATE;
3809 else if (bufp->regs_allocated == REGS_REALLOCATE)
3810 { /* Yes. If we need more elements than were already
3811 allocated, reallocate them. If we need fewer, just
3813 if (regs->num_regs < num_regs + 1)
3815 regs->num_regs = num_regs + 1;
3816 RETALLOC (regs->start, regs->num_regs, regoff_t);
3817 RETALLOC (regs->end, regs->num_regs, regoff_t);
3818 if (regs->start == NULL || regs->end == NULL)
3827 /* These braces fend off a "empty body in an else-statement"
3828 warning under GCC when assert expands to nothing. */
3829 assert (bufp->regs_allocated == REGS_FIXED);
3832 /* Convert the pointer data in `regstart' and `regend' to
3833 indices. Register zero has to be set differently,
3834 since we haven't kept track of any info for it. */
3835 if (regs->num_regs > 0)
3837 regs->start[0] = pos;
3838 regs->end[0] = (MATCHING_IN_FIRST_STRING
3839 ? ((regoff_t) (d - string1))
3840 : ((regoff_t) (d - string2 + size1)));
3843 /* Go through the first `min (num_regs, regs->num_regs)'
3844 registers, since that is all we initialized. */
3845 for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++)
3847 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
3848 regs->start[mcnt] = regs->end[mcnt] = -1;
3852 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
3854 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
3858 /* If the regs structure we return has more elements than
3859 were in the pattern, set the extra elements to -1. If
3860 we (re)allocated the registers, this is the case,
3861 because we always allocate enough to have at least one
3863 for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++)
3864 regs->start[mcnt] = regs->end[mcnt] = -1;
3865 } /* regs && !bufp->no_sub */
3867 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
3868 nfailure_points_pushed, nfailure_points_popped,
3869 nfailure_points_pushed - nfailure_points_popped);
3870 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
3872 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
3876 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
3882 /* Otherwise match next pattern command. */
3883 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3885 /* Ignore these. Used to ignore the n of succeed_n's which
3886 currently have n == 0. */
3888 DEBUG_PRINT1 ("EXECUTING no_op.\n");
3892 DEBUG_PRINT1 ("EXECUTING succeed.\n");
3895 /* Match the next n pattern characters exactly. The following
3896 byte in the pattern defines n, and the n bytes after that
3897 are the characters to match. */
3900 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
3902 /* This is written out as an if-else so we don't waste time
3903 testing `translate' inside the loop. */
3909 if (translate[(unsigned char) *d++] != (char) *p++)
3919 if (*d++ != (char) *p++) goto fail;
3923 SET_REGS_MATCHED ();
3927 /* Match any character except possibly a newline or a null. */
3929 DEBUG_PRINT1 ("EXECUTING anychar.\n");
3933 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
3934 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
3937 SET_REGS_MATCHED ();
3938 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
3946 register unsigned char c;
3947 boolean not = (re_opcode_t) *(p - 1) == charset_not;
3949 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
3952 c = TRANSLATE (*d); /* The character to match. */
3954 /* Cast to `unsigned' instead of `unsigned char' in case the
3955 bit list is a full 32 bytes long. */
3956 if (c < (unsigned) (*p * BYTEWIDTH)
3957 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
3962 if (!not) goto fail;
3964 SET_REGS_MATCHED ();
3970 /* The beginning of a group is represented by start_memory.
3971 The arguments are the register number in the next byte, and the
3972 number of groups inner to this one in the next. The text
3973 matched within the group is recorded (in the internal
3974 registers data structure) under the register number. */
3976 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
3978 /* Find out if this group can match the empty string. */
3979 p1 = p; /* To send to group_match_null_string_p. */
3981 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
3982 REG_MATCH_NULL_STRING_P (reg_info[*p])
3983 = group_match_null_string_p (&p1, pend, reg_info);
3985 /* Save the position in the string where we were the last time
3986 we were at this open-group operator in case the group is
3987 operated upon by a repetition operator, e.g., with `(a*)*b'
3988 against `ab'; then we want to ignore where we are now in
3989 the string in case this attempt to match fails. */
3990 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
3991 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
3993 DEBUG_PRINT2 (" old_regstart: %d\n",
3994 POINTER_TO_OFFSET (old_regstart[*p]));
3997 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
3999 IS_ACTIVE (reg_info[*p]) = 1;
4000 MATCHED_SOMETHING (reg_info[*p]) = 0;
4002 /* Clear this whenever we change the register activity status. */
4003 set_regs_matched_done = 0;
4005 /* This is the new highest active register. */
4006 highest_active_reg = *p;
4008 /* If nothing was active before, this is the new lowest active
4010 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4011 lowest_active_reg = *p;
4013 /* Move past the register number and inner group count. */
4015 just_past_start_mem = p;
4020 /* The stop_memory opcode represents the end of a group. Its
4021 arguments are the same as start_memory's: the register
4022 number, and the number of inner groups. */
4024 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4026 /* We need to save the string position the last time we were at
4027 this close-group operator in case the group is operated
4028 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4029 against `aba'; then we want to ignore where we are now in
4030 the string in case this attempt to match fails. */
4031 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4032 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4034 DEBUG_PRINT2 (" old_regend: %d\n",
4035 POINTER_TO_OFFSET (old_regend[*p]));
4038 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4040 /* This register isn't active anymore. */
4041 IS_ACTIVE (reg_info[*p]) = 0;
4043 /* Clear this whenever we change the register activity status. */
4044 set_regs_matched_done = 0;
4046 /* If this was the only register active, nothing is active
4048 if (lowest_active_reg == highest_active_reg)
4050 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4051 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4054 { /* We must scan for the new highest active register, since
4055 it isn't necessarily one less than now: consider
4056 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4057 new highest active register is 1. */
4058 unsigned char r = *p - 1;
4059 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4062 /* If we end up at register zero, that means that we saved
4063 the registers as the result of an `on_failure_jump', not
4064 a `start_memory', and we jumped to past the innermost
4065 `stop_memory'. For example, in ((.)*) we save
4066 registers 1 and 2 as a result of the *, but when we pop
4067 back to the second ), we are at the stop_memory 1.
4068 Thus, nothing is active. */
4071 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4072 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4075 highest_active_reg = r;
4078 /* If just failed to match something this time around with a
4079 group that's operated on by a repetition operator, try to
4080 force exit from the ``loop'', and restore the register
4081 information for this group that we had before trying this
4083 if ((!MATCHED_SOMETHING (reg_info[*p])
4084 || just_past_start_mem == p - 1)
4087 boolean is_a_jump_n = false;
4091 switch ((re_opcode_t) *p1++)
4095 case pop_failure_jump:
4096 case maybe_pop_jump:
4098 case dummy_failure_jump:
4099 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4109 /* If the next operation is a jump backwards in the pattern
4110 to an on_failure_jump right before the start_memory
4111 corresponding to this stop_memory, exit from the loop
4112 by forcing a failure after pushing on the stack the
4113 on_failure_jump's jump in the pattern, and d. */
4114 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4115 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4117 /* If this group ever matched anything, then restore
4118 what its registers were before trying this last
4119 failed match, e.g., with `(a*)*b' against `ab' for
4120 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4121 against `aba' for regend[3].
4123 Also restore the registers for inner groups for,
4124 e.g., `((a*)(b*))*' against `aba' (register 3 would
4125 otherwise get trashed). */
4127 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4131 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4133 /* Restore this and inner groups' (if any) registers. */
4134 for (r = *p; r < *p + *(p + 1); r++)
4136 regstart[r] = old_regstart[r];
4138 /* xx why this test? */
4139 if (old_regend[r] >= regstart[r])
4140 regend[r] = old_regend[r];
4144 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4145 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4151 /* Move past the register number and the inner group count. */
4156 /* \<digit> has been turned into a `duplicate' command which is
4157 followed by the numeric value of <digit> as the register number. */
4160 register const char *d2, *dend2;
4161 int regno = *p++; /* Get which register to match against. */
4162 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4164 /* Can't back reference a group which we've never matched. */
4165 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4168 /* Where in input to try to start matching. */
4169 d2 = regstart[regno];
4171 /* Where to stop matching; if both the place to start and
4172 the place to stop matching are in the same string, then
4173 set to the place to stop, otherwise, for now have to use
4174 the end of the first string. */
4176 dend2 = ((FIRST_STRING_P (regstart[regno])
4177 == FIRST_STRING_P (regend[regno]))
4178 ? regend[regno] : end_match_1);
4181 /* If necessary, advance to next segment in register
4185 if (dend2 == end_match_2) break;
4186 if (dend2 == regend[regno]) break;
4188 /* End of string1 => advance to string2. */
4190 dend2 = regend[regno];
4192 /* At end of register contents => success */
4193 if (d2 == dend2) break;
4195 /* If necessary, advance to next segment in data. */
4198 /* How many characters left in this segment to match. */
4201 /* Want how many consecutive characters we can match in
4202 one shot, so, if necessary, adjust the count. */
4203 if (mcnt > dend2 - d2)
4206 /* Compare that many; failure if mismatch, else move
4209 ? bcmp_translate (d, d2, mcnt, translate)
4210 : bcmp (d, d2, mcnt))
4212 d += mcnt, d2 += mcnt;
4214 /* Do this because we've match some characters. */
4215 SET_REGS_MATCHED ();
4221 /* begline matches the empty string at the beginning of the string
4222 (unless `not_bol' is set in `bufp'), and, if
4223 `newline_anchor' is set, after newlines. */
4225 DEBUG_PRINT1 ("EXECUTING begline.\n");
4227 if (AT_STRINGS_BEG (d))
4229 if (!bufp->not_bol) break;
4231 else if (d[-1] == '\n' && bufp->newline_anchor)
4235 /* In all other cases, we fail. */
4239 /* endline is the dual of begline. */
4241 DEBUG_PRINT1 ("EXECUTING endline.\n");
4243 if (AT_STRINGS_END (d))
4245 if (!bufp->not_eol) break;
4248 /* We have to ``prefetch'' the next character. */
4249 else if ((d == end1 ? *string2 : *d) == '\n'
4250 && bufp->newline_anchor)
4257 /* Match at the very beginning of the data. */
4259 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4260 if (AT_STRINGS_BEG (d))
4265 /* Match at the very end of the data. */
4267 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4268 if (AT_STRINGS_END (d))
4273 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4274 pushes NULL as the value for the string on the stack. Then
4275 `pop_failure_point' will keep the current value for the
4276 string, instead of restoring it. To see why, consider
4277 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4278 then the . fails against the \n. But the next thing we want
4279 to do is match the \n against the \n; if we restored the
4280 string value, we would be back at the foo.
4282 Because this is used only in specific cases, we don't need to
4283 check all the things that `on_failure_jump' does, to make
4284 sure the right things get saved on the stack. Hence we don't
4285 share its code. The only reason to push anything on the
4286 stack at all is that otherwise we would have to change
4287 `anychar's code to do something besides goto fail in this
4288 case; that seems worse than this. */
4289 case on_failure_keep_string_jump:
4290 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4292 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4293 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4295 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4299 /* Uses of on_failure_jump:
4301 Each alternative starts with an on_failure_jump that points
4302 to the beginning of the next alternative. Each alternative
4303 except the last ends with a jump that in effect jumps past
4304 the rest of the alternatives. (They really jump to the
4305 ending jump of the following alternative, because tensioning
4306 these jumps is a hassle.)
4308 Repeats start with an on_failure_jump that points past both
4309 the repetition text and either the following jump or
4310 pop_failure_jump back to this on_failure_jump. */
4311 case on_failure_jump:
4313 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4315 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4316 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4318 /* If this on_failure_jump comes right before a group (i.e.,
4319 the original * applied to a group), save the information
4320 for that group and all inner ones, so that if we fail back
4321 to this point, the group's information will be correct.
4322 For example, in \(a*\)*\1, we need the preceding group,
4323 and in \(\(a*\)b*\)\2, we need the inner group. */
4325 /* We can't use `p' to check ahead because we push
4326 a failure point to `p + mcnt' after we do this. */
4329 /* We need to skip no_op's before we look for the
4330 start_memory in case this on_failure_jump is happening as
4331 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4333 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4336 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4338 /* We have a new highest active register now. This will
4339 get reset at the start_memory we are about to get to,
4340 but we will have saved all the registers relevant to
4341 this repetition op, as described above. */
4342 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4343 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4344 lowest_active_reg = *(p1 + 1);
4347 DEBUG_PRINT1 (":\n");
4348 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4352 /* A smart repeat ends with `maybe_pop_jump'.
4353 We change it to either `pop_failure_jump' or `jump'. */
4354 case maybe_pop_jump:
4355 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4356 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4358 register unsigned char *p2 = p;
4360 /* Compare the beginning of the repeat with what in the
4361 pattern follows its end. If we can establish that there
4362 is nothing that they would both match, i.e., that we
4363 would have to backtrack because of (as in, e.g., `a*a')
4364 then we can change to pop_failure_jump, because we'll
4365 never have to backtrack.
4367 This is not true in the case of alternatives: in
4368 `(a|ab)*' we do need to backtrack to the `ab' alternative
4369 (e.g., if the string was `ab'). But instead of trying to
4370 detect that here, the alternative has put on a dummy
4371 failure point which is what we will end up popping. */
4373 /* Skip over open/close-group commands.
4374 If what follows this loop is a ...+ construct,
4375 look at what begins its body, since we will have to
4376 match at least one of that. */
4380 && ((re_opcode_t) *p2 == stop_memory
4381 || (re_opcode_t) *p2 == start_memory))
4383 else if (p2 + 6 < pend
4384 && (re_opcode_t) *p2 == dummy_failure_jump)
4391 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
4392 to the `maybe_finalize_jump' of this case. Examine what
4395 /* If we're at the end of the pattern, we can change. */
4398 /* Consider what happens when matching ":\(.*\)"
4399 against ":/". I don't really understand this code
4401 p[-3] = (unsigned char) pop_failure_jump;
4403 (" End of pattern: change to `pop_failure_jump'.\n");
4406 else if ((re_opcode_t) *p2 == exactn
4407 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
4409 register unsigned char c
4410 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4412 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
4414 p[-3] = (unsigned char) pop_failure_jump;
4415 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4419 else if ((re_opcode_t) p1[3] == charset
4420 || (re_opcode_t) p1[3] == charset_not)
4422 int not = (re_opcode_t) p1[3] == charset_not;
4424 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
4425 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4428 /* `not' is equal to 1 if c would match, which means
4429 that we can't change to pop_failure_jump. */
4432 p[-3] = (unsigned char) pop_failure_jump;
4433 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4437 else if ((re_opcode_t) *p2 == charset)
4440 register unsigned char c
4441 = *p2 == (unsigned char) endline ? '\n' : p2[2];
4444 if ((re_opcode_t) p1[3] == exactn
4445 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[4]
4446 && (p2[1 + p1[4] / BYTEWIDTH]
4447 & (1 << (p1[4] % BYTEWIDTH)))))
4449 p[-3] = (unsigned char) pop_failure_jump;
4450 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
4454 else if ((re_opcode_t) p1[3] == charset_not)
4457 /* We win if the charset_not inside the loop
4458 lists every character listed in the charset after. */
4459 for (idx = 0; idx < (int) p2[1]; idx++)
4460 if (! (p2[2 + idx] == 0
4461 || (idx < (int) p1[4]
4462 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
4467 p[-3] = (unsigned char) pop_failure_jump;
4468 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4471 else if ((re_opcode_t) p1[3] == charset)
4474 /* We win if the charset inside the loop
4475 has no overlap with the one after the loop. */
4477 idx < (int) p2[1] && idx < (int) p1[4];
4479 if ((p2[2 + idx] & p1[5 + idx]) != 0)
4482 if (idx == p2[1] || idx == p1[4])
4484 p[-3] = (unsigned char) pop_failure_jump;
4485 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
4490 p -= 2; /* Point at relative address again. */
4491 if ((re_opcode_t) p[-1] != pop_failure_jump)
4493 p[-1] = (unsigned char) jump;
4494 DEBUG_PRINT1 (" Match => jump.\n");
4495 goto unconditional_jump;
4497 /* Note fall through. */
4500 /* The end of a simple repeat has a pop_failure_jump back to
4501 its matching on_failure_jump, where the latter will push a
4502 failure point. The pop_failure_jump takes off failure
4503 points put on by this pop_failure_jump's matching
4504 on_failure_jump; we got through the pattern to here from the
4505 matching on_failure_jump, so didn't fail. */
4506 case pop_failure_jump:
4508 /* We need to pass separate storage for the lowest and
4509 highest registers, even though we don't care about the
4510 actual values. Otherwise, we will restore only one
4511 register from the stack, since lowest will == highest in
4512 `pop_failure_point'. */
4513 unsigned dummy_low_reg, dummy_high_reg;
4514 unsigned char *pdummy;
4517 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
4518 POP_FAILURE_POINT (sdummy, pdummy,
4519 dummy_low_reg, dummy_high_reg,
4520 reg_dummy, reg_dummy, reg_info_dummy);
4522 /* Note fall through. */
4525 /* Unconditionally jump (without popping any failure points). */
4528 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
4529 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
4530 p += mcnt; /* Do the jump. */
4531 DEBUG_PRINT2 ("(to 0x%x).\n", p);
4535 /* We need this opcode so we can detect where alternatives end
4536 in `group_match_null_string_p' et al. */
4538 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
4539 goto unconditional_jump;
4542 /* Normally, the on_failure_jump pushes a failure point, which
4543 then gets popped at pop_failure_jump. We will end up at
4544 pop_failure_jump, also, and with a pattern of, say, `a+', we
4545 are skipping over the on_failure_jump, so we have to push
4546 something meaningless for pop_failure_jump to pop. */
4547 case dummy_failure_jump:
4548 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
4549 /* It doesn't matter what we push for the string here. What
4550 the code at `fail' tests is the value for the pattern. */
4551 PUSH_FAILURE_POINT (0, 0, -2);
4552 goto unconditional_jump;
4555 /* At the end of an alternative, we need to push a dummy failure
4556 point in case we are followed by a `pop_failure_jump', because
4557 we don't want the failure point for the alternative to be
4558 popped. For example, matching `(a|ab)*' against `aab'
4559 requires that we match the `ab' alternative. */
4560 case push_dummy_failure:
4561 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
4562 /* See comments just above at `dummy_failure_jump' about the
4564 PUSH_FAILURE_POINT (0, 0, -2);
4567 /* Have to succeed matching what follows at least n times.
4568 After that, handle like `on_failure_jump'. */
4570 EXTRACT_NUMBER (mcnt, p + 2);
4571 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
4574 /* Originally, this is how many times we HAVE to succeed. */
4579 STORE_NUMBER_AND_INCR (p, mcnt);
4580 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt);
4584 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
4585 p[2] = (unsigned char) no_op;
4586 p[3] = (unsigned char) no_op;
4592 EXTRACT_NUMBER (mcnt, p + 2);
4593 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
4595 /* Originally, this is how many times we CAN jump. */
4599 STORE_NUMBER (p + 2, mcnt);
4600 goto unconditional_jump;
4602 /* If don't have to jump any more, skip over the rest of command. */
4609 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
4611 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4613 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4614 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
4615 STORE_NUMBER (p1, mcnt);
4620 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
4621 if (AT_WORD_BOUNDARY (d))
4626 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
4627 if (AT_WORD_BOUNDARY (d))
4632 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
4633 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
4638 DEBUG_PRINT1 ("EXECUTING wordend.\n");
4639 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
4640 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
4646 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
4647 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
4652 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
4653 if (PTR_CHAR_POS ((unsigned char *) d) != point)
4658 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
4659 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
4662 #if 0 /* not emacs19 */
4664 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
4665 if (PTR_CHAR_POS ((unsigned char *) d) + 1 != point)
4668 #endif /* not emacs19 */
4671 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
4676 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
4680 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
4682 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
4684 SET_REGS_MATCHED ();
4688 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
4690 goto matchnotsyntax;
4693 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
4697 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
4699 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
4701 SET_REGS_MATCHED ();
4704 #else /* not emacs */
4706 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
4708 if (!WORDCHAR_P (d))
4710 SET_REGS_MATCHED ();
4715 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
4719 SET_REGS_MATCHED ();
4722 #endif /* not emacs */
4727 continue; /* Successfully executed one pattern command; keep going. */
4730 /* We goto here if a matching operation fails. */
4732 if (!FAIL_STACK_EMPTY ())
4733 { /* A restart point is known. Restore to that state. */
4734 DEBUG_PRINT1 ("\nFAIL:\n");
4735 POP_FAILURE_POINT (d, p,
4736 lowest_active_reg, highest_active_reg,
4737 regstart, regend, reg_info);
4739 /* If this failure point is a dummy, try the next one. */
4743 /* If we failed to the end of the pattern, don't examine *p. */
4747 boolean is_a_jump_n = false;
4749 /* If failed to a backwards jump that's part of a repetition
4750 loop, need to pop this failure point and use the next one. */
4751 switch ((re_opcode_t) *p)
4755 case maybe_pop_jump:
4756 case pop_failure_jump:
4759 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4762 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
4764 && (re_opcode_t) *p1 == on_failure_jump))
4772 if (d >= string1 && d <= end1)
4776 break; /* Matching at this starting point really fails. */
4780 goto restore_best_regs;
4784 return -1; /* Failure to match. */
4787 /* Subroutine definitions for re_match_2. */
4790 /* We are passed P pointing to a register number after a start_memory.
4792 Return true if the pattern up to the corresponding stop_memory can
4793 match the empty string, and false otherwise.
4795 If we find the matching stop_memory, sets P to point to one past its number.
4796 Otherwise, sets P to an undefined byte less than or equal to END.
4798 We don't handle duplicates properly (yet). */
4801 group_match_null_string_p (p, end, reg_info)
4802 unsigned char **p, *end;
4803 register_info_type *reg_info;
4806 /* Point to after the args to the start_memory. */
4807 unsigned char *p1 = *p + 2;
4811 /* Skip over opcodes that can match nothing, and return true or
4812 false, as appropriate, when we get to one that can't, or to the
4813 matching stop_memory. */
4815 switch ((re_opcode_t) *p1)
4817 /* Could be either a loop or a series of alternatives. */
4818 case on_failure_jump:
4820 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4822 /* If the next operation is not a jump backwards in the
4827 /* Go through the on_failure_jumps of the alternatives,
4828 seeing if any of the alternatives cannot match nothing.
4829 The last alternative starts with only a jump,
4830 whereas the rest start with on_failure_jump and end
4831 with a jump, e.g., here is the pattern for `a|b|c':
4833 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
4834 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
4837 So, we have to first go through the first (n-1)
4838 alternatives and then deal with the last one separately. */
4841 /* Deal with the first (n-1) alternatives, which start
4842 with an on_failure_jump (see above) that jumps to right
4843 past a jump_past_alt. */
4845 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
4847 /* `mcnt' holds how many bytes long the alternative
4848 is, including the ending `jump_past_alt' and
4851 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
4855 /* Move to right after this alternative, including the
4859 /* Break if it's the beginning of an n-th alternative
4860 that doesn't begin with an on_failure_jump. */
4861 if ((re_opcode_t) *p1 != on_failure_jump)
4864 /* Still have to check that it's not an n-th
4865 alternative that starts with an on_failure_jump. */
4867 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4868 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
4870 /* Get to the beginning of the n-th alternative. */
4876 /* Deal with the last alternative: go back and get number
4877 of the `jump_past_alt' just before it. `mcnt' contains
4878 the length of the alternative. */
4879 EXTRACT_NUMBER (mcnt, p1 - 2);
4881 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
4884 p1 += mcnt; /* Get past the n-th alternative. */
4890 assert (p1[1] == **p);
4896 if (!common_op_match_null_string_p (&p1, end, reg_info))
4899 } /* while p1 < end */
4902 } /* group_match_null_string_p */
4905 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
4906 It expects P to be the first byte of a single alternative and END one
4907 byte past the last. The alternative can contain groups. */
4910 alt_match_null_string_p (p, end, reg_info)
4911 unsigned char *p, *end;
4912 register_info_type *reg_info;
4915 unsigned char *p1 = p;
4919 /* Skip over opcodes that can match nothing, and break when we get
4920 to one that can't. */
4922 switch ((re_opcode_t) *p1)
4925 case on_failure_jump:
4927 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4932 if (!common_op_match_null_string_p (&p1, end, reg_info))
4935 } /* while p1 < end */
4938 } /* alt_match_null_string_p */
4941 /* Deals with the ops common to group_match_null_string_p and
4942 alt_match_null_string_p.
4944 Sets P to one after the op and its arguments, if any. */
4947 common_op_match_null_string_p (p, end, reg_info)
4948 unsigned char **p, *end;
4949 register_info_type *reg_info;
4954 unsigned char *p1 = *p;
4956 switch ((re_opcode_t) *p1++)
4976 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
4977 ret = group_match_null_string_p (&p1, end, reg_info);
4979 /* Have to set this here in case we're checking a group which
4980 contains a group and a back reference to it. */
4982 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
4983 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
4989 /* If this is an optimized succeed_n for zero times, make the jump. */
4991 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4999 /* Get to the number of times to succeed. */
5001 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5006 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5014 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5022 /* All other opcodes mean we cannot match the empty string. */
5028 } /* common_op_match_null_string_p */
5031 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5032 bytes; nonzero otherwise. */
5035 bcmp_translate (s1, s2, len, translate)
5036 unsigned char *s1, *s2;
5040 register unsigned char *p1 = s1, *p2 = s2;
5043 if (translate[*p1++] != translate[*p2++]) return 1;
5049 /* Entry points for GNU code. */
5051 /* re_compile_pattern is the GNU regular expression compiler: it
5052 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5053 Returns 0 if the pattern was valid, otherwise an error string.
5055 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5056 are set in BUFP on entry.
5058 We call regex_compile to do the actual compilation. */
5061 re_compile_pattern (pattern, length, bufp)
5062 const char *pattern;
5064 struct re_pattern_buffer *bufp;
5068 /* GNU code is written to assume at least RE_NREGS registers will be set
5069 (and at least one extra will be -1). */
5070 bufp->regs_allocated = REGS_UNALLOCATED;
5072 /* And GNU code determines whether or not to get register information
5073 by passing null for the REGS argument to re_match, etc., not by
5077 /* Match anchors at newline. */
5078 bufp->newline_anchor = 1;
5080 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5084 return gettext (re_error_msgid[(int) ret]);
5087 /* Entry points compatible with 4.2 BSD regex library. We don't define
5088 them unless specifically requested. */
5090 #ifdef _REGEX_RE_COMP
5092 /* BSD has one and only one pattern buffer. */
5093 static struct re_pattern_buffer re_comp_buf;
5103 if (!re_comp_buf.buffer)
5104 return gettext ("No previous regular expression");
5108 if (!re_comp_buf.buffer)
5110 re_comp_buf.buffer = (unsigned char *) malloc (200);
5111 if (re_comp_buf.buffer == NULL)
5112 return gettext (re_error_msgid[(int) REG_ESPACE]);
5113 re_comp_buf.allocated = 200;
5115 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5116 if (re_comp_buf.fastmap == NULL)
5117 return gettext (re_error_msgid[(int) REG_ESPACE]);
5120 /* Since `re_exec' always passes NULL for the `regs' argument, we
5121 don't need to initialize the pattern buffer fields which affect it. */
5123 /* Match anchors at newlines. */
5124 re_comp_buf.newline_anchor = 1;
5126 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5131 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5132 return (char *) gettext (re_error_msgid[(int) ret]);
5140 const int len = strlen (s);
5142 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5144 #endif /* _REGEX_RE_COMP */
5146 /* POSIX.2 functions. Don't define these for Emacs. */
5150 /* regcomp takes a regular expression as a string and compiles it.
5152 PREG is a regex_t *. We do not expect any fields to be initialized,
5153 since POSIX says we shouldn't. Thus, we set
5155 `buffer' to the compiled pattern;
5156 `used' to the length of the compiled pattern;
5157 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
5158 REG_EXTENDED bit in CFLAGS is set; otherwise, to
5159 RE_SYNTAX_POSIX_BASIC;
5160 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
5161 `fastmap' and `fastmap_accurate' to zero;
5162 `re_nsub' to the number of subexpressions in PATTERN.
5164 PATTERN is the address of the pattern string.
5166 CFLAGS is a series of bits which affect compilation.
5168 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
5169 use POSIX basic syntax.
5171 If REG_NEWLINE is set, then . and [^...] don't match newline.
5172 Also, regexec will try a match beginning after every newline.
5174 If REG_ICASE is set, then we considers upper- and lowercase
5175 versions of letters to be equivalent when matching.
5177 If REG_NOSUB is set, then when PREG is passed to regexec, that
5178 routine will report only success or failure, and nothing about the
5181 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
5182 the return codes and their meanings.) */
5185 regcomp (preg, pattern, cflags)
5187 const char *pattern;
5192 = (cflags & REG_EXTENDED) ?
5193 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
5195 /* regex_compile will allocate the space for the compiled pattern. */
5197 preg->allocated = 0;
5200 /* Don't bother to use a fastmap when searching. This simplifies the
5201 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
5202 characters after newlines into the fastmap. This way, we just try
5206 if (cflags & REG_ICASE)
5210 preg->translate = (char *) malloc (CHAR_SET_SIZE);
5211 if (preg->translate == NULL)
5212 return (int) REG_ESPACE;
5214 /* Map uppercase characters to corresponding lowercase ones. */
5215 for (i = 0; i < CHAR_SET_SIZE; i++)
5216 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
5219 preg->translate = NULL;
5221 /* If REG_NEWLINE is set, newlines are treated differently. */
5222 if (cflags & REG_NEWLINE)
5223 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
5224 syntax &= ~RE_DOT_NEWLINE;
5225 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
5226 /* It also changes the matching behavior. */
5227 preg->newline_anchor = 1;
5230 preg->newline_anchor = 0;
5232 preg->no_sub = !!(cflags & REG_NOSUB);
5234 /* POSIX says a null character in the pattern terminates it, so we
5235 can use strlen here in compiling the pattern. */
5236 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
5238 /* POSIX doesn't distinguish between an unmatched open-group and an
5239 unmatched close-group: both are REG_EPAREN. */
5240 if (ret == REG_ERPAREN) ret = REG_EPAREN;
5246 /* regexec searches for a given pattern, specified by PREG, in the
5249 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
5250 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
5251 least NMATCH elements, and we set them to the offsets of the
5252 corresponding matched substrings.
5254 EFLAGS specifies `execution flags' which affect matching: if
5255 REG_NOTBOL is set, then ^ does not match at the beginning of the
5256 string; if REG_NOTEOL is set, then $ does not match at the end.
5258 We return 0 if we find a match and REG_NOMATCH if not. */
5261 regexec (preg, string, nmatch, pmatch, eflags)
5262 const regex_t *preg;
5265 regmatch_t pmatch[];
5269 struct re_registers regs;
5270 regex_t private_preg;
5271 int len = strlen (string);
5272 boolean want_reg_info = !preg->no_sub && nmatch > 0;
5274 private_preg = *preg;
5276 private_preg.not_bol = !!(eflags & REG_NOTBOL);
5277 private_preg.not_eol = !!(eflags & REG_NOTEOL);
5279 /* The user has told us exactly how many registers to return
5280 information about, via `nmatch'. We have to pass that on to the
5281 matching routines. */
5282 private_preg.regs_allocated = REGS_FIXED;
5286 regs.num_regs = nmatch;
5287 regs.start = TALLOC (nmatch, regoff_t);
5288 regs.end = TALLOC (nmatch, regoff_t);
5289 if (regs.start == NULL || regs.end == NULL)
5290 return (int) REG_NOMATCH;
5293 /* Perform the searching operation. */
5294 ret = re_search (&private_preg, string, len,
5295 /* start: */ 0, /* range: */ len,
5296 want_reg_info ? ®s : (struct re_registers *) 0);
5298 /* Copy the register information to the POSIX structure. */
5305 for (r = 0; r < nmatch; r++)
5307 pmatch[r].rm_so = regs.start[r];
5308 pmatch[r].rm_eo = regs.end[r];
5312 /* If we needed the temporary register info, free the space now. */
5317 /* We want zero return to mean success, unlike `re_search'. */
5318 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
5322 /* Returns a message corresponding to an error code, ERRCODE, returned
5323 from either regcomp or regexec. We don't use PREG here. */
5326 regerror (errcode, preg, errbuf, errbuf_size)
5328 const regex_t *preg;
5336 || errcode >= (sizeof (re_error_msgid) / sizeof (re_error_msgid[0])))
5337 /* Only error codes returned by the rest of the code should be passed
5338 to this routine. If we are given anything else, or if other regex
5339 code generates an invalid error code, then the program has a bug.
5340 Dump core so we can fix it. */
5343 msg = gettext (re_error_msgid[errcode]);
5345 msg_size = strlen (msg) + 1; /* Includes the null. */
5347 if (errbuf_size != 0)
5349 if (msg_size > errbuf_size)
5351 strncpy (errbuf, msg, errbuf_size - 1);
5352 errbuf[errbuf_size - 1] = 0;
5355 strcpy (errbuf, msg);
5362 /* Free dynamically allocated space used by PREG. */
5368 if (preg->buffer != NULL)
5369 free (preg->buffer);
5370 preg->buffer = NULL;
5372 preg->allocated = 0;
5375 if (preg->fastmap != NULL)
5376 free (preg->fastmap);
5377 preg->fastmap = NULL;
5378 preg->fastmap_accurate = 0;
5380 if (preg->translate != NULL)
5381 free (preg->translate);
5382 preg->translate = NULL;
5385 #endif /* not emacs */
5389 make-backup-files: t
5391 trim-versions-without-asking: nil
projects / gnulib.git / blob